---

[Quoting amb]

---

[Quoting awpagan]

how about signal one side, power and ground the other?

---

It's almost that way now. Virtually all ground is on the top side. The power traces are on the bottom because otherwise the ground traces would get in the way.

---

I see..

two 4700uF might be overkill yes, but this is diy.
How about breaking the track between the 2 caps for a CRC?

The main goal was to keep the actual audio signal path as short as possible since this amp has such high bandwidth.

If you look at the schematic you can see that it is drawn with the input transistors at the left, current mirrors on top/bottom, the voltage sources to the right of the input bjts and then the output stage.

But, the board does not have this layout. Looking at just one channel (see amb's latest image), the input transistors Q2 and Q3 are in the middle of the layout. This puts them very close to Q1 and Q4. Since the current mirrors formed by Q1/Q7 and Q4/Q8 are part of the signal path (and part of the gain), Q1/Q7 and Q4/Q8 are also placed very close together. Finally, the driver transistors Q9 and Q11 are situated very close to the devices that drive them, namely Q9 and Q11.

Looking at the board the signal paths from Q2->Q1->Q7->Q9 and Q3->Q4->Q8->Q11 are very tight with minimal capacitance.

Unlike the schematic, the voltage source components (Q5, Q6, capacitors and resistors) are located on the other side of the input devices. These components perform a different function in the signal path and can be situated away from the Q2-->Q9 and Q3-->Q11 sequences.

My original board design had this layout with the signal devices very close together. AMB's evolution of the board put the devices even closer together and in alignment.

Thus, the gain/power portion of each channel is entirely contained in the space between the input headers and the output transistors. This is a little less than two inches and should enable the design to perform at its maximum bandwidth.

[Edited by runeight on 11-11-2006 at 03:18 AM.]

---

[Quoting awpagan]

two 4700uF might be overkill yes, but this is diy.

---

---

[Quote]

How about breaking the track between the 2 caps for a CRC?

---

Going too large before and after reg's seem to defeat the purpose of the reg.
better to have the ground leg of the 317/337/78/79's isolated from large cap pulses.

This link maybe helpful for inplementation of lm3*7 / 78/79's
http://www.diyau...18336#post18336

[Edited by awpagan on 11-11-2006 at 08:25 AM.]

Here are my thoughts on the capacitor sizes in the PS, remembering that this is diy.

First, I like to have as little ripple as is reasonable at the input to the regs. Of course, these regulators have good ripple rejection, but the more ripple you have at the input the more work the active circuitry in the reg has to do continuously in response. And this work generates some amount of junk on the rails. I broke the capacitor into two separate caps for two reasons: 1) so you can mix and match a little more with different sizes and 2) to make the space on the board a little more flexible.

The second reason for the large capacitance is to give the regs a large reservoir to work from when current demands are placed on the PS.

The 2x4700u could be too much, but because of the board design, you can play with this within some range of values to suite your own tastes.

The cheap reg ICs like the LM317 don't have very good ripple rejection at high frequencies so I think you should add a polypropylene or polyester capacitor after the main DC smoothing caps to filter noise out.

---

[Quote]

The second reason for the large capacitance is to give the regs a large reservoir to work from when current demands are placed on the PS.

---

[Edited by Dave_M on 11-11-2006 at 03:12 PM.]

If the amp were generating a 30mA peak sine wave then the amp has to pull 30mA from the PS on one cycle and then return 30mA on the other half. The PS has to be able to source and sink the currents. Now music isn't a sine wave, but it does have this behavior nonetheless. The regulator's ability to source/sink current depends on what's in front of it.

But, I can't argue with cap sizes. You guys can try whatever you think is appropriate.

Dave_M, there is provision on the board for 220u electrolytic and 2u2 film cap after the regulator but not before it.

[Edited by runeight on 11-11-2006 at 04:56 PM.]

[Edited by Dave_M on 11-11-2006 at 05:57 PM.]

The schematic also includes the alternate servo component values.

[Edited by runeight on 11-13-2006 at 12:18 AM.]

This substitution has no real downside to it. The upside is that it gives the amp a very high input impedance. And so we are going to make this the standard configuration.

Schematics are attached.

Edit: there was an error on the voltages for C3, C4, C12, C13. I've replaced the schematics with ones showing the correct voltages.

[Edited by runeight on 11-16-2006 at 12:52 AM.]

This is to let everyone know that the interest check thread is still ongoing, and I will be running simulations of this amp very soon (runeight already did his own sims). I will be ordering boards after that. I will report on the progress each step of the way.

Also, I would think it might make sense to add protection diodes to the 7815/7915 regs. Finally, is there any specific reason you chose the triangular pin configuration for the TO-92 parts? I think the straight tape configuration would make it a bit less tight in many areas, and would definitly make it much easier to use either CBE or EBC parts, depending on what people already have or can get easily.

---

[Quoting amb]

---

[Quoting awpagan]

How about breaking the track between the 2 caps for a CRC?

---

I doubt that it would make any improvement and would be throwing extra parts at the circuit for no reason.

---

Why 0.5ohm emitter resistors of the output CFPs? Is this just to achieve low output impedance, at the expense of some protection?

Finally, and this is just a technicality, but you cant really call the ground a star ground, can you? I thought to be a true star ground, everything was required to have its own return to the star point.

[Edited by cetoole on 11-16-2006 at 07:52 AM.]

Although we like the jfets as the stock configuration, because the jfets and bjts are pin compatible, it is easy to build the all bjt version using the resistor values from previous schematics. In the event that the jfets are hard to get. Performance will be about the same with lower Zi. But, you have a good point about their availability.

0.5R for lower Zo and because the amp tends to have lower distortion with these smaller value resistors. And since there is no NFB we have to reduce THD using what's available.

The PSRR seems to be about 40db from each rail.

I guess for the star ground it depends on what you mean by everything. If you mean that every single component that is grounded has a separate ground run, then I don't think this is what is meant by a star ground. Generally the circuit is star grounded in blocks. That is, a PS block will have one connection to the star ground, the input section another, and so forth. In this PCB design, each PS half has one connection to the star, the opamp servo section has one connection, the output has one connection. The only place where "starness" is slightly violated is the amp itself where the input ground and the internal amp grounds are connected together before they are attached to the star. But, this is ok as these ground references should be to the same place.

How much is the input impedance on the amps that me and dBel84 have made? We are talking tens of kΩ, right? How about adding local feedback to get the THD down? That would be an improvement wouldn't it? I am not someone who opposes change, but these modifications are changing some of the main features in the design of the original amp such as no NFB, all discrete BJTs, unregulated PSU.

I'm not saying you are making the amp worse, just that it now deserves a name change. IMHO of course!

---

[Quote]

I'm not saying you are making the amp worse, just that it now deserves a name change. IMHO of course!

---

I was thinking the same thing.

I'm happy to change the name if you all think that's appropriate. In the beginning I wanted to be sure to preserve Benny's original contribution to the amp's design. The amp is still Benny's basic topology of all discrete, complementary, no NFB, etc. But the mods are getting to be significant and maybe Benny should no longer be held responsible for what this amp does.

---

[Quoting Dave_M]

How much is the input impedance on the amps that me and dBel84 have made? We are talking tens of kΩ, right?

---

This depends on the vol pot, of course, but leaving this out of the picture let's assume that each version has its input grounded through 1M. This is probably too high for the bjt version, but for comparison sake we'll use it.

In this case:

BJT Input Zi = 100k
JFET Input Zi = 925k

So, if you hang a 100k vol pot onto the front of the bjt version under these circumstances you'll have a Zi of 50k. IF you use a leak resistor of 100k from the input to ground as well as a vol pot then you'll have a Zi of 33k. You don't absolutely need this resistor but there is space for it on the board and it will protect the amp when the vol pot is not connected.

If you hang 100k vol pot onto the front of the jfet version you'll have a Zi of slightly less than 100k. Most of this from the pot. In this case there is no reason not to put the 1M leak resistor on the board since it hardly affects the Zi.

[Edited by runeight on 11-16-2006 at 01:46 PM.]

But before you make the boards for this amp, you need to design 2 brand new amps, CKI and CKII first

[Edited by Dave_M on 11-16-2006 at 01:48 PM.]