Yes, more current gives better drive but also more gain so you have to change the values of the RIAA components.
If it is not a topology problem i have no clue. Maybe the single RIAA simply does not work and we need two.
The Paradise runs on 10mA in the input and there is ample overload margin.
Does your simulation show that clipping behavior ?
If it is not a topology problem i have no clue. Maybe the single RIAA simply does not work and we need two.
The Paradise runs on 10mA in the input and there is ample overload margin.
Does your simulation show that clipping behavior ?
No, simulation doesn't show any clipping.
However IMHO 10mA might not be enough to have good high frequency margine.
It is a matter to establish during design specification what is the max frequency you want to be able to bring the output at clipping at.
If we set a full swing at 25KHz for example I suspect that more than 10mA will be required.
Anyway, simulating more, circled out what the problem is with the mismatching of the two sides.
On the bench I was noticing that mismatching of the two halves gets bigger and bigger as frequency increses and I couldn't explain why.
Here is circuit simulated using devices on the left side with different Hfe and shows exactly the behaviour I am seeing in real life (I thought it was pretty cool).
The conclusion is that tight matching for this topology is a MUST!
Here is the simulation shot.
Attachments
Yes, tight matching is necessary. The problem increases in the treble because a small phase delay difference has much stronger effect in the treble then it has in the bass.
That is true for common mode distortion too.
Maybe better matching solves the clipping.
For the here and now you can try to trim the emitter resistors in the current mirror.
That is true for common mode distortion too.
Maybe better matching solves the clipping.
For the here and now you can try to trim the emitter resistors in the current mirror.
yeah I fully agree with you.
I would say that it is to a good point, however needs more fine tuning.
I also agree that strange clipping might improve significantly with tight matching.
Here is the updated circuit using Celvin's buffer (however I haven't tried it yet I am going to do a shot out of output stage on this next prototype I am going to run).
I also added the option of doubling up the input devices don't think it is necessary per se, but current going through the device (about 8mA) is a little high to handle for the input device chosen and also linearity goes a bit down too.
Attachments
I agrre scaling the mirror is not good. I did a power amp where I did scaling,(input to VAS) it did not do me any good, then i changed the mirror and had three parallel devices in the out-side then problems was gone.... maybe that will be the solution here also..I don't believe that it is a question of mismatch, more that your mirror transistors at not working at the same point on the curve.
You removed the cascode in the buffer..??
You removed the cascode in the buffer..??
That creates a miss match. I learned from Keantoken that the quasi saturation region is critical in current mirrors. That is the region under 0.7V where BJTs are not full on but conducting on a sliding curve. Scaling the mirror shifts that curve.
The solution with multiple output is good i think.
I have seen that in IC opamps so it is a mature measure.
The solution with multiple output is good i think.
I have seen that in IC opamps so it is a mature measure.
I thought of that, just didn't think it could actually create a problem on the mirror side riscaling it.
Let me rework the schematic to work about 30-35mA
You are correct, these are just place holders.
I am currently using LSK from linear system, I go by the top of my head 312 and 352 I believe.
I do really like their sound!
The PNP :
http://www.linearsystems.com/assets/media/file/datasheets/LS350-2.pdf
They look good but can only stand 10mA maximum. I am surprised, that is a really low number.
Ft, Hfe and Cob are very nice.
http://www.linearsystems.com/assets/media/file/datasheets/LS350-2.pdf
They look good but can only stand 10mA maximum. I am surprised, that is a really low number.
Ft, Hfe and Cob are very nice.
Yes the device I am referring is that one!
I don't know how the new schematic will work out.
I will start drawing the layout in these coming days.
I am listening in the background to the current version.
It sounds amazing.
Unfortunately there is this distortion especially at low frequency when there area peaks of power.
I don't know how to capture this situation on the bench.
Basically I am listening to ray brown on solar energy.
When there are strong transition it clips or distort and it is not the cartridge.
I can't figure out what the cause is.
Beside that the sound of this thing is pretty amazing, powerful opens refined and so on.
I will prepare the new layout.
Any of the developer guys here would like to receive a proto board for testing as well?
I think that it would be extremely useful to test drive this baby from different viewpoints
I don't know how the new schematic will work out.
I will start drawing the layout in these coming days.
I am listening in the background to the current version.
It sounds amazing.
Unfortunately there is this distortion especially at low frequency when there area peaks of power.
I don't know how to capture this situation on the bench.
Basically I am listening to ray brown on solar energy.
When there are strong transition it clips or distort and it is not the cartridge.
I can't figure out what the cause is.
Beside that the sound of this thing is pretty amazing, powerful opens refined and so on.
I will prepare the new layout.
Any of the developer guys here would like to receive a proto board for testing as well?
I think that it would be extremely useful to test drive this baby from different viewpoints
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.