Thanks to everyone who has shown interest in the amp. I see that Digi is getting some takers for the board and we are still working with him to finalize the board and to make its part numbers agree with the schematics that will be in the new SOHA article.

We can start some new conversations on this thread, including using discrete buffers for the output stage. I'll try to start this when I have a little more time.

In the meantime, Digi01 can you post another schematic of your drop-in discrete buffer. I think we can make a start with that.

Some questions for members who have build the standard SOHA 'point to point':

- when you build your amp up on a breadboard (and not yet in the case) how close could you place the trafo near the ampsections before you get audible (or irritating) hum because of the strayfield of the trafo?

- when you cased all sections did you encounter hum problems from the trafo? How did you solve that problems?

- was it in your case needed to place the trafo outside the amp case?

- Did someone looked with a scope to see if there were too much interference between trafo and ampsections? At wich distance did you reach a acceptable hum/noise level?

I am asking this because I personally find it very irritating when hum is interfering with music at low level. I have read that in other headampprojects the advise is given to place a trafo (especially non-toiroids) in a separate little box placed a couple of feet away from the amp.

So what are the expierences with the SOHA on this topic. With that knowledge we can learn how to plan the best lay out when casing up.

Pieter.

The reason I did not have ordered some pcb’s yet is that I think I have not yet fully (in-)sight into the whole picture of the SOHA with respect to the schematic that will be used for the pcb and I have not yet insight in the full layout of the definitive pcb.

I hope Digi01 will postpone the deadline for the ordering a little so everyone can have fully insight before definitive ordering. Doing so everyone has the opportunity to comment on the lay out. Also I expect that when we have fully insight there will be provoked some extra demand for the pcb’s. By that alone it is quite reasonable that the 100+ border will be passed, so that there are economical benefits for us all to be gained.

I think we are yet in the position to ‘let’s make things better’ (I am not a stakeholder of Philips!), but never mind:

- My goal with comments is to give some remarks so that we all get the most optimised euro card sized pcb for a optimal functioning SOHA. I do not expect that such a SOHA will outperform everything (it can not by any means because of structural reasons, but for DIY non-experienced builders of tube based amps it can offer very much, I expect myself!) but I only hope that within its design goals and using the right components it can perform in a very musical way. That alone is quite an accomplishment and that can be for some of us a ‘stepping stone’ to even higher quality tube based amps.

- The lay out for the pcb I have seen till today looks good and much more flexible than before , but in my view there are some little aspects that can make it even better.

- Until now in the supply for the opamp there is room for using little regulators 78Lx/79Lx with 100ma capacity. Is it possible to use also 78/79 TO220 components? Or are the mounting holes for the little one’s already suitable for the TO220 one’s? I am asking this because of the fact that the pssr figures of the TO220 regulators are far better, some 30db better is the case! I think that the sonically results of amps are better when pssr is lower. May be I am wrong, let me know.

- For the filament supply there is used a lm337 based supply. This is very simply implemented. When you look into the spec sheets there are some little measures you can take to diminish the ripple from the supply with very little addition of components. So a very cost effective solution is within reach to severely diminish noise. Why not implemented on the pcb? Look at the spec sheets and also at the schematic for the Tread from Tangent, only a couple of components extra gives much better results, on the pcb there is enough room for them.

- When a discrete buffer is used there will be extra load on the standard 200ma supply configuration (look into the remarks of Runeight in SOHA ONE). This means that when the pcb will contain some discrete buffer option the power supply configuration has to be changed to a level by which the current demand can be fulfilled. This can provoke some different heat dissipation patterns other than the smart solution now found in the ‘standard supply’ for a SOHA. I think there should be some additional thought on positioning components that are to be expected to get more heat dissipation than in the standard configuration. I think of some of the ccs components and the regulators especially, because higher spec trafo or beefed up supply leads to higher dc level and consequentional higher heat dissipation in some spots of the amp. Using higher spec components (fe TO220 regulators) in combination with enough room for clip-on heat sinks can - may be - solve this (possible) problem. Correct me please if I am wrong.

- I hope that the implementation of some sort of discrete buffer on the pcb will also make it possible that one can use them into Class A level (to resonable extend that is). That means that there will be more heat produced at some spots in the buffer. In the Millet Hybrid case (a drop-in solution) there was too little space for a heat sink supported solution (nevertheless the buffer boards for the Millet are very successful, no misunderstanding about that!). I hope that on the SOHA pcb there will be enough space to implement enough clip-on heat sinking for the buffer transistors.

- I hope that on relevant spots there will be test points projected, because otherwise you have a problem. In any way, trimming plate voltages and testing B+ voltage should be easy to do from above the pcb, that’s only to do with test points from above. If not than you have to disconnect the amp board every time you have to trim some voltages I am afraid. This is in the contrary to point to point because there the accessibility of component traces is a whole other story.

Well this is it for the moment. May be I have more to comment on when the whole story of the SOHA is published on Headwize and the buffer type/pcb board is shown in its more definitive form.

Pieter.

On this schematic, the CCSs are shown as PCB mounting holes. The values for all components will be shown in the article in tables, but they have already been posted on the various schematics before. I will enclose a CCS jumpering/component table later.

Perhaps this will help some.

Your questions and points are very reasonable and would make the board much more flexible. It is already more flexible now with the changes that Digi is making based on this schematic. I think that Digi will have to comment on his ability to include additional features. The only change to the schematic that I can see from your comments would be a small electrolytic at the control pin for the LM337.

But, if you would like to take this schematic and mark it up, I would like to see what other changes you are proposing. Now would be the time to do this.

[Edited by runeight on 05-21-2006 at 06:02 PM.]

---

[Quote]

- when you build your amp up on a breadboard (and not yet in the case) how close could you place the trafo near the ampsections before you get audible (or irritating) hum because of the strayfield of the trafo?

---

---

[Quote]

- when you cased all sections did you encounter hum problems from the trafo? How did you solve that problems?

---

---

[Quote]

- was it in your case needed to place the trafo outside the amp case?

---

---

[Quote]

So what are the expierences with the SOHA on this topic. With that knowledge we can learn how to plan the best lay out when casing up.

---

[Edited by digi01 on 07-29-2006 at 03:40 AM.]

---

[Quoting Neurotica]

It's my personal belief that when mounting the trafo internally as I have done that a toroidal unit is the only way to go. I don't have any evidence to back up that claim, it's merely a guess from reading opinions of others in various forums and articles. I will be trying out a toroidal unit in the next few days so I'll post my results.

---

Yes, I'd be interested in that route as well. Although I've been contemplating mounting the trafo on the top of the case in a cover of some sort, similar to top mounting a trafo on a typical tube amp. I would like to find a regular low voltage center tapped filament supply that I could use for the purpose so I could mount the thing on top and just have a small hole that the wires run through, but the only ones I've seen that are close are 28VCT and 36VCT.

Thanks.

Zang

- Until now in the supply for the opamp there is room for using little regulators 78Lx/79Lx with 100ma capacity. Is it possible to use also 78/79 TO220 components? Or are the mounting holes for the little one’s already suitable for the TO220 one’s? I am asking this because of the fact that the pssr figures of the TO220 regulators are far better, some 30db better is the case! I think that the sonically results of amps are better when pssr is lower. May be I am wrong, let me know.

///I will try to dig out some rooms for to220 packaged regulators

- For the filament supply there is used a lm337 based supply. This is very simply implemented. When you look into the spec sheets there are some little measures you can take to diminish the ripple from the supply with very little addition of components. So a very cost effective solution is within reach to severely diminish noise. Why not implemented on the pcb? Look at the spec sheets and also at the schematic for the Tread from Tangent, only a couple of components extra gives much better results, on the pcb there is enough room for them.

///
I know idea,I have a single unit regulator PCB design has implemented it and it work fine.but for SOHA,I wish simple is the best.
P.S.
In personal point,I like AC powered filament much more.

- When a discrete buffer is used there will be extra load on the standard 200ma supply configuration (look into the remarks of Runeight in SOHA ONE). This means that when the pcb will contain some discrete buffer option the power supply configuration has to be changed to a level by which the current demand can be fulfilled. This can provoke some different heat dissipation patterns other than the smart solution now found in the ‘standard supply’ for a SOHA. I think there should be some additional thought on positioning components that are to be expected to get more heat dissipation than in the standard configuration. I think of some of the ccs components and the regulators especially, because higher spec trafo or beefed up supply leads to higher dc level and consequentional higher heat dissipation in some spots of the amp. Using higher spec components (fe TO220 regulators) in combination with enough room for clip-on heat sinks can - may be - solve this (possible) problem. Correct me please if I am wrong.

///
the buffer I am going to make is class AB based.To92 pcakaged regulator would handle it fine.

- I hope that the implementation of some sort of discrete buffer on the pcb will also make it possible that one can use them into Class A level (to resonable extend that is). That means that there will be more heat produced at some spots in the buffer. In the Millet Hybrid case (a drop-in solution) there was too little space for a heat sink supported solution (nevertheless the buffer boards for the Millet are very successful, no misunderstanding about that!). I hope that on the SOHA pcb there will be enough space to implement enough clip-on heat sinking for the buffer transistors.

///
I am interested in class A buffer.I will consider it carefully.thank you!

- I hope that on relevant spots there will be test points projected, because otherwise you have a problem. In any way, trimming plate voltages and testing B+ voltage should be easy to do from above the pcb, that’s only to do with test points from above. If not than you have to disconnect the amp board every time you have to trim some voltages I am afraid. This is in the contrary to point to point because there the accessibility of component traces is a whole other story.

///
yeah,good ideaI have set up some test point at the PCB.I will show all the details tomorrow.

Zang

[Edited by digi01 on 05-22-2006 at 02:51 AM.]

http://headwize...id=6609&fpage=3

[Edited by raromachine on 05-22-2006 at 03:29 AM.]

Let me first say that I think you guys are really interested in the process of ‘how can we get the best out of the board for ‘our’ version of SOHA.

After studying your reactions and looking into the latest schematic I come to the following remarks. May be you can use them.

I could not use your suggestion to comment in the drawing of the schematic itself because I am not familiar with how to do that. Instead I have written down my comment and refer to some component places and numbers. I hope that makes it all clear.

I like the idea that you are going to implement TO220 packages for VR1 and VR2. Only question is: have both kind of packages the same pin out? If they are the same than implementation is quite easy, otherwise rather complex. If the pin outs are different I would advise you only to use TO220 on the pcb. They are easy to get, not pricey and gives better pssr.

AC or DC filament supply? I know that ‘real tube diehards’ always advocate that AC supply for filament is sonically ‘better’ than DC (as with LM337 in the SOHA). But how to get the correct and continuous 12.6 V from a trafo that also has some other functions? However when cost is less restricted (we are not dealing with a maxed out SOHA at this moment) one would like to use a separate trafo and make it possible to use AC or DC with the choice to the builder him/herself. I have looked into some Dutch tube amp projects and have seen that there both ways of supplying are used with good results. There is one point in which they all are remarkably the same and that is the use of extensive smoothing capacity. More than 10.000uf is ‘normal’ especially at the place of C12 on the SOHA board. Always the primary goal was for them to get very very low ripple.

It depends on you definitive layout plans if you are going to make more room for C12. If you make more room than a couple of C12’s in parallel can make up for a very good starting point for the filament supply. When we want to stay very practical with respect to SOHA: when I look at the last schematic and the last version of the pcb I see that there is made room for rather big versions for C12 and C11. So only change here something when you are going to make more structural changes. See below.

Is it still the goal to mount the trafo on the euro board? I for myself are not going to do that because of the risk of a too high hum level (with low level signals) just because of the stray field of the trafo. Looking into the experiences of some of the point to point builders I am not taking the risk. I think I will look for a little separate case for the trafo or I make some sort of my own wallmart, that is a abs plastic box with socket pins and enough room to mount a appropriate print trafo.

If it is not the goal to mount the trafo on board than there is quite some room available. That room could be used for spreading out some of the components, putting on some extra caps and there will be plenty of room for a strategic placement of some kind of buffer (if necessary with enough room for heat sinking for trannies that get hot because of running in Class A). Moving all caps to the right makes it possible to place the buffer near the tube so short signal traces can be used.

How hot can the LND150 and twin J113 became when B+ is rather high(-er) than 60V as specified? When really hot than there should be enough place for clip on heat sinks, just in case they get too hot and collapse. I know there are types for that, only I do not know their sizes. For the LND150 there is always enough room because you use only one, only the distance wrt the trimpot can be a problem. For the twin J113 it is another case, may be when you turn them a little and so that they are opposite to each other than there is enough room for clip on heatsinks. There has to be enough room with respect to the trimpots.

Another point is : separate (little) trafo’s for each function and voltage/current demand? That could be a interesting experiment that could be even quite cheap, especially when you use some surplus printtrafo’s or something like that. Any buffer type makes it necessary to use a little higher current type than standard prescribed and I can not foresee how the voltage equilibrium will react to that. Too high current in relation to a too low load will give more than 60V at the B+ (I assume). Downstream you have to dissipate some more V into heat so that there is max 40V left. Can the standard components withstand that? I think we should hold several option open. May be it is practical to put a jumper in the trace/line just before C5 and C6. May be also between D4 and input VR3. By doing so you can build the standard supply or you can build your own (separate) supply configuration. You also avoid that a builder has to cut the trace himself, that is not a neat solution.

By the way: has someone ever contemplated the opportunity to use a CLC in the B+ instead of the standard CRC? I understand it when you have a strict budget for the SOHA, but maybe it could be build sometime to see if it matters. In many higher quality tube amps there is preference for CLC at that point. Given the fact that the ECC82 tube is used at very low ma and voltage it seems to me that a high quality very clean and stable B+ is quite relevant, but at this moment it is quite out of the scope. Does your simulations of the SOHA gives some perspective here? Or are the CCS’s more than adequate at this point?

I have seen again that the buffer of Digi01 (see posted one) is quite another thing than the one from Runeight (see SOHA ONE post 737). Also there is the broadly documented and thoroughly tested experience from Steinchen/MilletHybridgroup (see DIYFORUMS.ORG and/or his personal site on this topic) that has lead to a discrete buffer that runs in class A a long way and is a direct drop in for a chip buffer (replaces buff 634). I hope that the real experts are going to join us in this faze, because we can not have all of the types. Or is there a universal flexible to use configuration for pinholes on the pcb in which only the number of stages and different components has to be used? Or is it so that you are planning a drop in version in place of the standard output chip? Or will there be a separate place on the board and can you choose one or the other by a toggle switch? Here there is in my opinion still a moment for making some decisions before closing the development of the pcb for the SOHA. You see also here I am on slippery ice, so let the real experts help us out so we can make the right choices.

In the standard SOHA there is no output cap because the output chip does not need it. When a buffer will be used at the output: is there still no need for a output cap? No one wants to put his/her phone’s at any risk. Or should you use some extra functionality on the board to avoid problems (there is room on the euro sized board when no on board trafo is used)? Only questions, I really do not know if there is any risk, but than you have only to reassure me.

With respect to the implementation of the LM337 my advise is that you put a extra component from the adj pin to ground. This gives 17db extra ripple rejection according to page two of the spec sheet from NatSemi. On the spec sheet there is advised to use a 10uf capacitor (solid tantalum approx 25V or higher voltage) (or a 100uf electrolytic: see the site of TANGENT AUDIO schematic Tread or Steps).

I am looking forward to see where you placed the test points.

So this is it for now, much success in deciding on the definitive lay out for the SOHA board. I will following the process closely and with positive interest.

Pieter.

First a description of the trafo/tube geography. My trafo is approx. 3 inches to the left of my tube which is elevated 2 inches off the RS protoboard. The input wires are shielded but the tube grids are not. All ground wires are routed towards the front and the star ground in located in a far corner away from both trafo and tube.

Second, I have no output resistors installed.

Third, I have yet to box it up but it will be encased in a custom hard maple box.

Now to my results of hum tests...

1. First, I only get hum in my right channel, which is the one closest to the trafo. The left channel is absolutely hum free as far as my ears can hear.

2. I used tightly twisted plenum CAT5 for the heater wires set at right angles and I get insignificant hum in the right channel with 300ohm phones and no output resistors. With 32ohm, it is another story but the hum is not overwhelming just an annoyance.

3. I have also found that my hum is tube dependant. For example, I get more hum in from a GE 5963 in comparison to a RCA (Siemens) 12au7.

4. I created more hum when I breadboarded the protoboard to a sheet of aluminum. Even though I connected the aluminum to ground the hum increased in the right channel and in the left. That's why the case is going to be all wood and I'll pray for no sparks.

5. Grounded shield (aluminum sheet and aluminum foil) attempts all failed in regards to ridding the right channel of all hum. Therefore, I am going to live with it and start enjoying the sweet sounding amp.

Lastly - My question to digi01...

In your prototype build on pcb how much hum (if any) are you getting?

I built a SOHA and it worked fine (no hum) on UK mains volts.
I've shipped it to runeight (after modifying for 115V) and now, apparently, it has slight hum in the left channel and slight buzz in the right channel, but not enough to be annoying.
Hmmmm (no pun intended)...

I used 150Ω resistors at the outputs so that Grado SR-60 and Sennheiser HD-600 are the same volume and neither exhibited hum that I could discern.

But on US mains, clearly there's a issue.

the best remedy is, as amb pointed out several times, to place the transformer as far asway from the tube as possible. Next step is to use a well filtered and regulated power supply.

[Edited by Steinchen on 05-22-2006 at 03:47 PM.]

---

[Quoting Steinchen]

aluminum doesn't work as a shield, particulary not against the magnetic field of a transformer, you need MU-metal or steel (or copper)

---

Steinchen, I've thought of taking the recommended PC mount trafo and putting it on it's own little bit of protoboard. Mounting the board to the top of the enclosure on standoffs with the wires running through a grommetted hole down into the chassis. Then put a steel cover with some top side vents over the trafo to cover it's butt uglyness up. Do you think this method might shield the magnetic field? This would still allow the trafo to be one with the main amp chassis as well.

spec modification below:
1,volume,combo footprint for alps 27 and standard pot
2,the footprint of output jack is remove for easy front panel process
3,test points added
4,flexile caps footprint,lead space 5/7.5mm,18mm diameter
5,coupling caps,lead space 5/7.5/15mm
6,VR1,VR2 reseted for to92/to220 packaged regulator
6,LM337 move away from the caps for hot release
7,final PCB size is 100mm x 119mm
8,parts number updated
9,added 2 holes for stack classA buffer
thanks Alex,Bill and Pieter.

P.S.
the pcb is double layer design,so I can use breakaway grounding concept.the power ground and signal ground is 'separate' without loops.there are big ground plate at the top layer for signal ground shielding.the ground plate was set up with gridding for physical stabilization.

best,
Zang

EDIT:attachment replaced.

[Edited by digi01 on 05-23-2006 at 03:48 AM.]

---

[Quote]

I like the idea that you are going to implement TO220 packages for VR1 and VR2. Only question is: have both kind of packages the same pin out?

---

yes,they are the same pin out.

---

[Quote]

AC or DC filament supply? I know that ‘real tube diehards’ always advocate that AC supply for filament is sonically ‘better’ than DC

---

Yeah,I think SOHA sould be a DIY friendly designso let's make the things easy.

---

[Quote]

Is it still the goal to mount the trafo on the euro board? I for myself are not going to do that because of the risk of a too high hum level

---

I have the same plan.the pcb size is 10mm x 119m,we have 100mm x 41mm room leave.

Also, while the lm337 has better psrr if it has a cap on the adjustment pin, this is not likely to help too much as the noise on the heater circuit is probably not a problem. Still, if there is room on the board, why not?

to alex and pieter,
I will add a cap on the adjustment pin as option,there is enough room

- first the name: you have to remove one l in the word amplifier.
- now you have to change the name anyway: I think you also could add version number and year of production. You have a little space there.
- 'standard pot', does that pot has the same pin footprint as alp 27? I have seen mostely they differ a little, not an item when you use an offboard one.
- testpoints looks very good to me.
- may be it is optical on my computerscreen but it looks like the width size of the vertical couplingcap is a little smaller than the horizontal one. At least lines looks like that. When there is enough fysical room for standard seizes with RM15 it is ok. But only you have the exact mm seizes.
- on the last schematic from Alex I see near the ccs and just before the couplingcap a triangle 's p c on board jumper'. May be I am not looking good enough, but I could not discouver such spots on your pcb yet.

What kind of issue is there with the tubesocket? I missed that.

Will there be a short testtraject on the latest version of the pcb, you know 'the devil is always in the detail'?

I hope we are in the final drawing stage now!

Pieter.