Dear folks,
please excuse my lame attempt at a big-mouthed name for a simply updated version of the '88 Cello Encore amplifier that sold for something like 10000$ back then.
The amp features an Jfet input differential (LTP), balanced VAS with current mirror (and with what Self calls a beta enhancer) and a driver stage with bootstrapped source followers. A lot of fine details to be found there.
Next, it is a no negative feedback design.
All the connections you see going from the VAS back to the LTP are frequency compensation and not loop feedback as you can see from the use of the tiny pF-caps being in series. These caps become a short only for very high frequencies well above the audible band (several hundred kHz). Further this special arrangement of frequency compensation, a loop over LTP and VAS (e.g. C15/R55) with individual compensation for the VAS and LTP (the resistor/cap series like C6/R38 or C5/R33), shows very close similarity to Bob Cordells Mosfet amp published in AES '84 (Figure 8) which features a stunning 300V/mus slew rate.
Further details on the topology were discussed in the original thread
http://www.diyaudio.com/forums/showthread.php?s=&threadid=116237&highlight=
with the original schematic from
http://www.nasotec.co.kr/acecart/bin/boarddata/DataRoom/114.pdf
Coming to what I did: the original old parts like the 2N5415 can be found even still today, however they're just old parts, easily superseded by new parts at a fraction of their price (think of Jocko Homos new law!). The choice of the small signal transistors like the beta-enhancer Q31/Q32 or the LTP-cascode Q26/Q27 is not critical, use whatever you have. As you see I have the nice 2SC2705/2SA1145 at hand.
For the LTP VBE current source transistor I chose a low-noise transistor, the 2SC1844, to improve noise. However that's again not strictly necessary, but won't hurt.
For the VAS-current mirror I chose Sanyos 2SC2911 simply because I need a slightly higher power dissipation part here and I like the symmetry of having TO-126 parts in the VAS and in the current mirror. If I had 2SD669As I would use them here.
The critical part for stability in this amp is the VAS, especially the VAS-transistors Q20/Q21. Their Cob limits intrinsically the speed of the VAS and building on top of this further stabilisation is provided by R33/C5. Mess that balance up and be prepared to redo the whole frequency compensation.
The original 2N5416 has a Cob of 25pF, so I chose Hitachis 2SB649A with the same capacitance. I'm aware that this is already an obsolete part, so you have following alternatives: use a different bjt with that Cob (like Toshibas 2SC5171) or Miller-Cdom it to reach that value.
The constant current diodes J508 X3/X4 are purely there for simulation, you can of course use the original CR220 as well or do it like me and use 2SK170 with a resistor to do the job.
The voltage clamp (Baker clamp, R30/D4/X1/C3/...)) is in normal operation completely inactive, so no fancy parts here.
The MJE15030/MJE15031 used for driver stage and VBE-multiplier I replaced with the supposedly more complementary parts MJE15034/MJE15035, but you can use the original parts if you wish. Compensation formed by C18/C20 should not be affected.
Outputs are also in my eyes not critical, I will use the nice MJW0281A/MJW0302A pair here, but you can use the original parts MJ15024/MJ15025, or the more modern parts MJL21193/MJL21194 which are more similar to the original parts, but feature plastic cases.
Keep in mind, it's beneficial to use complementary matched drivers and outputs (as in the original schematic) to give a balanced load to the VAS.
I didn't draw a PSU, as I think everybody interested will cook his own anyway. Since the original rails are not known, I chose about +-35V to get the spec'd output power of 50W/8R, 100W/4R. Unless you use a regulated PSU this voltage will vary anyway.
Short comment on the schematic: Resistors R57/R58 should be a 50k trimmer for adjusting offset, RT is also a trimmer to adjust bias of the output stage.
At last I want to say thank you to some people that helped me during this fun: thank you Steve for the detailed analysis and private discussion! A kind thanks to Pedro as well for kindly providing me with the nice parts and further thank you very much Prof. Leach for discussing with me the frequency compensation.
Of course thank you very much Nelson Pass, without your generous dedication to DIY I would never have come so far.
So guys, I'm happy about every comment!
Have fun, Hannes
please excuse my lame attempt at a big-mouthed name for a simply updated version of the '88 Cello Encore amplifier that sold for something like 10000$ back then.
The amp features an Jfet input differential (LTP), balanced VAS with current mirror (and with what Self calls a beta enhancer) and a driver stage with bootstrapped source followers. A lot of fine details to be found there.
Next, it is a no negative feedback design.
All the connections you see going from the VAS back to the LTP are frequency compensation and not loop feedback as you can see from the use of the tiny pF-caps being in series. These caps become a short only for very high frequencies well above the audible band (several hundred kHz). Further this special arrangement of frequency compensation, a loop over LTP and VAS (e.g. C15/R55) with individual compensation for the VAS and LTP (the resistor/cap series like C6/R38 or C5/R33), shows very close similarity to Bob Cordells Mosfet amp published in AES '84 (Figure 8) which features a stunning 300V/mus slew rate.
Further details on the topology were discussed in the original thread
http://www.diyaudio.com/forums/showthread.php?s=&threadid=116237&highlight=
with the original schematic from
http://www.nasotec.co.kr/acecart/bin/boarddata/DataRoom/114.pdf
Coming to what I did: the original old parts like the 2N5415 can be found even still today, however they're just old parts, easily superseded by new parts at a fraction of their price (think of Jocko Homos new law!). The choice of the small signal transistors like the beta-enhancer Q31/Q32 or the LTP-cascode Q26/Q27 is not critical, use whatever you have. As you see I have the nice 2SC2705/2SA1145 at hand.
For the LTP VBE current source transistor I chose a low-noise transistor, the 2SC1844, to improve noise. However that's again not strictly necessary, but won't hurt.
For the VAS-current mirror I chose Sanyos 2SC2911 simply because I need a slightly higher power dissipation part here and I like the symmetry of having TO-126 parts in the VAS and in the current mirror. If I had 2SD669As I would use them here.
The critical part for stability in this amp is the VAS, especially the VAS-transistors Q20/Q21. Their Cob limits intrinsically the speed of the VAS and building on top of this further stabilisation is provided by R33/C5. Mess that balance up and be prepared to redo the whole frequency compensation.
The original 2N5416 has a Cob of 25pF, so I chose Hitachis 2SB649A with the same capacitance. I'm aware that this is already an obsolete part, so you have following alternatives: use a different bjt with that Cob (like Toshibas 2SC5171) or Miller-Cdom it to reach that value.
The constant current diodes J508 X3/X4 are purely there for simulation, you can of course use the original CR220 as well or do it like me and use 2SK170 with a resistor to do the job.
The voltage clamp (Baker clamp, R30/D4/X1/C3/...)) is in normal operation completely inactive, so no fancy parts here.
The MJE15030/MJE15031 used for driver stage and VBE-multiplier I replaced with the supposedly more complementary parts MJE15034/MJE15035, but you can use the original parts if you wish. Compensation formed by C18/C20 should not be affected.
Outputs are also in my eyes not critical, I will use the nice MJW0281A/MJW0302A pair here, but you can use the original parts MJ15024/MJ15025, or the more modern parts MJL21193/MJL21194 which are more similar to the original parts, but feature plastic cases.
Keep in mind, it's beneficial to use complementary matched drivers and outputs (as in the original schematic) to give a balanced load to the VAS.
I didn't draw a PSU, as I think everybody interested will cook his own anyway. Since the original rails are not known, I chose about +-35V to get the spec'd output power of 50W/8R, 100W/4R. Unless you use a regulated PSU this voltage will vary anyway.
Short comment on the schematic: Resistors R57/R58 should be a 50k trimmer for adjusting offset, RT is also a trimmer to adjust bias of the output stage.
At last I want to say thank you to some people that helped me during this fun: thank you Steve for the detailed analysis and private discussion! A kind thanks to Pedro as well for kindly providing me with the nice parts and further thank you very much Prof. Leach for discussing with me the frequency compensation.
Of course thank you very much Nelson Pass, without your generous dedication to DIY I would never have come so far.
So guys, I'm happy about every comment!
Have fun, Hannes
h_a said:
I agree with you.
If keeping the original drivers and output transistors, we will do very well.
John Curl in past has used MJ15024/MJ15025 in some of his power amps.
These have a bit higher FT (4MHz) than MJ15003/4
But above all, MJ15024 in TO-3 case, will take most heat well.
MJE15031/30 are the favourite of many high end amplifier designers!
Now, somewhere in this forum, is an old post, where one guy enlightened me,
that MJE15028/MJE15029 pair is BETTER to use than MJE15030 MJE15031.
(if needed C-E voltage is not above rating)
It is apparent that not only the MAX voltage rating is different here.
My own Spice simulations verify this.
They do NOT use same spice model.
----------------------
From the clear and nice PDF schematic, I can see that this is
A SERIOUS Amplifier
Details are designed with care.
Nothing that somebody has put together in a just a coffeebreak pause.
Great Project!
Thanks.
llineup - friendly hifi regards
Hi lineup!
Oh that's unfortunately not a firm confirmation. I just recently came across bjts sharing the same die but a different package - both also had completely different spice-models. If you want I can try to dig the reference!
In what regard? The reason why I chose the MJE15034/MJE15035 is that somebody measured them and confirmed that they're well complementary (at least beta and fT):
http://www.diyaudio.com/forums/showthread.php?postid=1502185#post1502185
You're talking about the original amp, right? Yes a nice amp! It is said to be Colangelos work (he founded Cello with Mark Levinson), though I think John Curl thought it was Dick Burwens work. Well, maybe somebody knows more on this. Anyway, it's a nice design!
Thanks for your reply!
Have fun, Hannes
Oh that's unfortunately not a firm confirmation. I just recently came across bjts sharing the same die but a different package - both also had completely different spice-models. If you want I can try to dig the reference!
In what regard? The reason why I chose the MJE15034/MJE15035 is that somebody measured them and confirmed that they're well complementary (at least beta and fT):
http://www.diyaudio.com/forums/showthread.php?postid=1502185#post1502185
You're talking about the original amp, right? Yes a nice amp! It is said to be Colangelos work (he founded Cello with Mark Levinson), though I think John Curl thought it was Dick Burwens work. Well, maybe somebody knows more on this. Anyway, it's a nice design!
Thanks for your reply!
Have fun, Hannes
Hi Lumba!
Sure I keep the U403, as this amp already has a bit too much gain for my needs! (28.5dB for unbalanced input!) It anyway becomes even worse with the new parts that again feature more gain.
What do you dislike about the 2SB649A? (I'm only talking about genuine Hitachi parts of course.)
Ok there are better ones, but IMHO it's a nice part nevertheless. And I would like to avoid artificial Miller-compensation.
But of course there's still room left in the schematic and on the pcb
Have fun, Hannes
Sure I keep the U403, as this amp already has a bit too much gain for my needs! (28.5dB for unbalanced input!) It anyway becomes even worse with the new parts that again feature more gain.
What do you dislike about the 2SB649A? (I'm only talking about genuine Hitachi parts of course.)
Ok there are better ones, but IMHO it's a nice part nevertheless. And I would like to avoid artificial Miller-compensation.
But of course there's still room left in the schematic and on the pcb
Have fun, Hannes
Marcel gave me the U403 some time ago and said that they're more linear than 2SK389. I guess he should know since his old employee used them in some specialized equipment. Well anyway, if not better I'm satisfied even with the specs of the original Cello Encore which sported 0.3% THD+N at full power.
Your point about current vs. voltage amplification of the 2SB649A completely escapes me?
Lumba, you're sometimes hard to understand, maybe you could give me another hint?
Have fun, Hannes
Your point about current vs. voltage amplification of the 2SB649A completely escapes me?
Lumba, you're sometimes hard to understand, maybe you could give me another hint?
Have fun, Hannes
Hannes,
transistors are optimized for a large variety of applications and have accordingly different properties. Choosing suitable types is a basic condition for achieving desired results.
Unfortunatelly, I am unable to elucidate it better like this and can unlikely make usable contributions in this case.
transistors are optimized for a large variety of applications and have accordingly different properties. Choosing suitable types is a basic condition for achieving desired results.
Unfortunatelly, I am unable to elucidate it better like this and can unlikely make usable contributions in this case.
Hi lineup!
I just recently came across bjts sharing the same die but a different package - both also had completely different spice-models. If you want I can try to dig the reference!
Yeah, I believe you, h_a.
Same transistor can also have diffferent spice models.
In what regard?
The reason why I chose the MJE15034/MJE15035 is that somebody measured them and confirmed that they're well complementary
(at least beta and fT):
http://www.diyaudio.com/forums/showthread.php?postid=1502185#post1502185
Thanks for this information. Of course you should use those.
Your infomant may be just as right, as is mine.
Now, if could remember the reason why MJE15028/29 is a better choice, than MJE15030/31.
But I can not. Maybe a bit higher gain .. maybe better complementary Vbe .. I do not know.
I did search our forum for this old 'MJE15028 post'. But haven't found it, yet.
I did find this however:
---------------
When I refer to the original amplifier circuit, and if anyone is using the original circuit,
I mean that we should think twice, before exchange the Default and very good devices.
It is not as simple as you take a circuit designed and trimmed for one kind of transistor
and put in a 'better modern device'.
Good amplifier designers will change the currents/resistors/capacitors
to fit optimally for a special setup of transistors.
So many times when unexperieced diy-ers think they will get a better ampllfier
by change to new 'better' output or driver transistors, with no further trimming,
'the improvement' will be just an illusion in their mind.
You think what you want to think.
I know this is not the case with your version, Hannes.
You have actually tried to optimize the setup to your chosen devices.
---------------
Thanks for your reply!
Have fun, Hannes
Same to you, Hannes.
When somebody does the effort to tell us about a new amp and start a new topic
I find it common courtesy to at least give a sign we have read this material.
Most people post, Because they have something good to tell others.
I will follow this topic, and if I find I have some constructive input to make,
I will post here again.
... several minds together, can do more than one mind,
no matter how genious one head can be.
lineup regards
I just recently came across bjts sharing the same die but a different package - both also had completely different spice-models. If you want I can try to dig the reference!
Yeah, I believe you, h_a.
Same transistor can also have diffferent spice models.
In what regard?
The reason why I chose the MJE15034/MJE15035 is that somebody measured them and confirmed that they're well complementary
(at least beta and fT):
http://www.diyaudio.com/forums/showthread.php?postid=1502185#post1502185
Thanks for this information. Of course you should use those.
Your infomant may be just as right, as is mine.
Now, if could remember the reason why MJE15028/29 is a better choice, than MJE15030/31.
But I can not. Maybe a bit higher gain .. maybe better complementary Vbe .. I do not know.
I did search our forum for this old 'MJE15028 post'. But haven't found it, yet.
I did find this however:
Nelson Pass said:
---------------
When I refer to the original amplifier circuit, and if anyone is using the original circuit,
I mean that we should think twice, before exchange the Default and very good devices.
It is not as simple as you take a circuit designed and trimmed for one kind of transistor
and put in a 'better modern device'.
Good amplifier designers will change the currents/resistors/capacitors
to fit optimally for a special setup of transistors.
So many times when unexperieced diy-ers think they will get a better ampllfier
by change to new 'better' output or driver transistors, with no further trimming,
'the improvement' will be just an illusion in their mind.
You think what you want to think.
I know this is not the case with your version, Hannes.
You have actually tried to optimize the setup to your chosen devices.
---------------
Thanks for your reply!
Have fun, Hannes
Same to you, Hannes.When somebody does the effort to tell us about a new amp and start a new topic
I find it common courtesy to at least give a sign we have read this material.
Most people post, Because they have something good to tell others.
I will follow this topic, and if I find I have some constructive input to make,
I will post here again.
... several minds together, can do more than one mind,
no matter how genious one head can be.
lineup regards
Hi lineup, thank you for your kind words!
That's a very noble attitude! Of course you're a gentleman of the old school.
That's exactly what I also had in mind
@HKC: thanks a lot for your kind words as well!
Sure. For unbalanced connections tie it to ground.
Honestly I still don't understand why such a low input impedance was chosen for the - input (5k) which makes it a bit on the edge of being usefull (especially with high output impedance preamps, which could approach roughly 1k!).
It can only have something todo with the frequency compensation loop that is at two different points connected to the inverting side of the differential.
Anyway, if you decide to use it only unbalanced, you could make a permanent connection of -in to ground.
Have fun, Hannes
That's a very noble attitude! Of course you're a gentleman of the old school.
That's exactly what I also had in mind
@HKC: thanks a lot for your kind words as well!
Sure. For unbalanced connections tie it to ground.
Honestly I still don't understand why such a low input impedance was chosen for the - input (5k) which makes it a bit on the edge of being usefull (especially with high output impedance preamps, which could approach roughly 1k!).
It can only have something todo with the frequency compensation loop that is at two different points connected to the inverting side of the differential.
Anyway, if you decide to use it only unbalanced, you could make a permanent connection of -in to ground.
Have fun, Hannes
h_a said:Hi lineup, thank you for your kind words!
That's a very noble attitude! Of course you're a gentleman of the old school.
That's exactly what I also had in mind
@HKC: thanks a lot for your kind words as well!
Sure. For unbalanced connections tie it to ground.
Honestly I still don't understand why such a low input impedance was chosen for the - input (5k) which makes it a bit on the edge of being usefull (especially with high output impedance preamps, which could approach roughly 1k!).
It can only have something todo with the frequency compensation loop that is at two different points connected to the inverting side of the differential.
Anyway, if you decide to use it only unbalanced, you could make a permanent connection of -in to ground.
Have fun, Hannes
Hi Hannes
Thank you for your prompt reply.
I have been seeking a good balance but small output power amp I guess this is the power amp I need.
If possible, would you draw and post a balanced schematic for this power amp? Because I am using a great balance pre-amp I built few months ago which is Borbely's design.
As to the input impedance I think I will try both 5K & 1K to see what is the difference.
In addition, I noted most of the original parts are available from Mouser except the input JFets U403 which I will use 2N5566 instead and CR220 I will use sorted JFet to substitute.
Best Regards
Hi HKC,
no need to draw a balanced schematic! For balanced connections the - pin of the symmetrical connection is connected to -in of the amp. I didn't state it before because I thought that was clear.
Please note the input impedance of the amp is fixed at 5k for -in!
I'm sure that building the original design will give you great results!
By the way, you don't need sorted jfets to replace the CR220. You just need to find the right resistor for the used jfet (between gate and source) You could even use a pot as a simple solution.
All the best, Hannes
no need to draw a balanced schematic! For balanced connections the - pin of the symmetrical connection is connected to -in of the amp. I didn't state it before because I thought that was clear.
Please note the input impedance of the amp is fixed at 5k for -in!
I'm sure that building the original design will give you great results!
By the way, you don't need sorted jfets to replace the CR220. You just need to find the right resistor for the used jfet (between gate and source)
You could even use a pot as a simple solution.All the best, Hannes
h_a said:Hi HKC,
no need to draw a balanced schematic! For balanced connections the - pin of the symmetrical connection is connected to -in of the amp. I didn't state it before because I thought that was clear.
Please note the input impedance of the amp is fixed at 5k for -in!
I'm sure that building the original design will give you great results!
By the way, you don't need sorted jfets to replace the CR220. You just need to find the right resistor for the used jfet (between gate and source)
All the best, Hannes
Hi Hannes
Thank you again for the information you given. They are very helpful and useful for this Cello power amp.
By the way where did you get the JFet U403? I have checked with Mouser, RS & Fernell but none of them have it.
Best Regards
The PCB
I was just asked about my pcb-layout, so I thought I post a short info for those interested.
The pcb-layout is nearly finished and I hope to get it done soon.
It's a very compact layout* - 12x18 cm if I recall correctly -, double sided (without vias!), featuring star grounding instead of a ground plane to avoid parasitics (just look at the pF-caps, you know what I mean), large trace widths and large trace distances (much larger than required for professional production).
Further it allows as much as possible to use the original parts or completely different parts with additional Miller-compensation (VAS).
In case somebody wants a set, he or she is welcome to order them with me.
However please understand that I don't want to give away Gerber-files.
This is just a short info on what is coming, there's by no means a hurry.
Have fun, Hannes
*the size is set by the enclosure I want to use
I was just asked about my pcb-layout, so I thought I post a short info for those interested.
The pcb-layout is nearly finished and I hope to get it done soon.
It's a very compact layout* - 12x18 cm if I recall correctly -, double sided (without vias!), featuring star grounding instead of a ground plane to avoid parasitics (just look at the pF-caps, you know what I mean), large trace widths and large trace distances (much larger than required for professional production).
Further it allows as much as possible to use the original parts or completely different parts with additional Miller-compensation (VAS).
In case somebody wants a set, he or she is welcome to order them with me.
However please understand that I don't want to give away Gerber-files.
This is just a short info on what is coming, there's by no means a hurry.
Have fun, Hannes
*the size is set by the enclosure I want to use
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.