ha!
Papa spoiled all the fun .....
Not for me.... you and mikegranger may carry on, cause I still don't understand the finesse of this design.
Papa says: " The 385's allow proper operation into low impedances." 😕
How?
Also: "The diodes protect the output Gates and give symmetric clipping."😕
How?
So if somebody can explain in more detail, I'll be happy.
Yes, I have the schematic printed out and it's laying on the couch next to me, last two days 😀
let me paraphrase what I already wrote ( as answer to same question ) on Baby DiyA :
consider that we have rails of +/-22Vdc
that means (not counting losses precisely) that we have 44Vpp envelope for output swing
amp's gain is 5V/V (autoformer ratio)
so , for full blast on output we need 44Vpp/50= 8.8Vpp at input
that's 8.8/2/1.41= 3.11Vrms
so , if there is any idiot available ( and believe me , it always is ) , he certainly will bang full blast of something into amp's input
in that case , output node (point where diodes are connected ) will try to go both to roof and cellar , resulting in heavy clipping and most probably even more damage
so , just imagine effect of clamping diodes (D1-D2-D3) - they are clamping output node to rails , limiting max . signal amplitude
NB that output node is in strong correlation with mosfet gates , so clamping of outout is in fact achieved with gate clamping
Q6 and Q7 are there to maintain no more than 1V25 across corresponding resistors , saving LED from nasty peaks and maintaining smooth biasing
btw. did you put second schematic , framed , on the wall ?
consider that we have rails of +/-22Vdc
that means (not counting losses precisely) that we have 44Vpp envelope for output swing
amp's gain is 5V/V (autoformer ratio)
so , for full blast on output we need 44Vpp/50= 8.8Vpp at input
that's 8.8/2/1.41= 3.11Vrms
so , if there is any idiot available ( and believe me , it always is ) , he certainly will bang full blast of something into amp's input
in that case , output node (point where diodes are connected ) will try to go both to roof and cellar , resulting in heavy clipping and most probably even more damage
so , just imagine effect of clamping diodes (D1-D2-D3) - they are clamping output node to rails , limiting max . signal amplitude
NB that output node is in strong correlation with mosfet gates , so clamping of outout is in fact achieved with gate clamping
Q6 and Q7 are there to maintain no more than 1V25 across corresponding resistors , saving LED from nasty peaks and maintaining smooth biasing
btw. did you put second schematic , framed , on the wall ?
Last edited:
let me paraphrase what I already wrote
amp's gain is 5V/V (autoformer ratio)
so , for full blast on output we need 44Vpp/50= 8.8Vpp at input
that's 8.8/2/1.41= 3.11Vrms
so , if there is any idiot available ( and believe me , it always is ) , he certainly will bang full blast of something into amp's input
in that case , output node (point where diodes are connected ) will try to go both to roof and cellar , resulting in heavy clipping and most probably even more damage
so , just imagine effect of clamping diodes (D1-D2-D3) - they are clamping output node to rails , limiting max . signal amplitude
NB that output node is in strong correlation with mosfet gates , so clamping of outout is in fact achieved with gate clamping
Q6 and Q7 are there to maintain no more than 1V25 across corresponding resistors , saving LED from nasty peaks and maintaining smooth biasing
btw. did you put second schematic , framed , on the wall ?
Thanks Zen Mod..... That helps me a lot. I didn't think about the input transformer not having an inherent voltage limit as most active stages would have. WalterW - I don't pretend that I fully understood even after Papa gave his explanation, so I'm glad you asked for more information. I was guessing the diodes were there for clamping voltage coming from the output rather than the input, so except for being completely wrong I was getting closer to the true purpose. I had no clue as to why the 1.2 volt clamps on the output though. This has been really fun learning puzzle!
I should have said the D1 - D3 give symmetric clipping into low impedance
loads. They also protect the Mosfets from input transients.
More interesting are the roles of the LM385's - their voltage is slightly higher
than the forward voltage of the opto inputs. Think about what the output
currents and Vgs figures might look like in action.
😎
loads. They also protect the Mosfets from input transients.
More interesting are the roles of the LM385's - their voltage is slightly higher
than the forward voltage of the opto inputs. Think about what the output
currents and Vgs figures might look like in action.
😎
yup
exact situation why I almost melted what's left of what eons ago was brilliant mind .....
Pa gave me the puzzle , and I was furious trying to decipher what those diodes (D1-D3) are for ....... (pretty standard case of not seeing Tree from the Wood)
when I cried oomphteenth time, he responded with pretty much same sentence
that was first time when I seriously fired up LTSpice (being ignorant Greedy Boy with napkin and calculator) and then I simply couldn't confirm heavy load thesis
what I confirmed with sims , really not in a mood later to confirm in vivo with my precioussss Papa's Koan M2 , is this*** :
D1-D3 are guarding those fragile gates in case of idiotic sine driving them (load value pretty irrelevant) clamping them to similar values above positive and negative rails , also resulting in symmetrical clipping
LM385 are saving opto LED of nasty current spikes , in all big current sucking conditions , be it moderate voltage/heavy load or heavy voltage/moderate load
disclaimer - sim is sim (especially when made by Mighty ZM) , and Mithrandir is Mithrandir****
***I'm lazy to repeat and to post screenshots
****(finding great fun not just pulling someone else's leg , but even his own , here and there)

exact situation why I almost melted what's left of what eons ago was brilliant mind .....

Pa gave me the puzzle , and I was furious trying to decipher what those diodes (D1-D3) are for ....... (pretty standard case of not seeing Tree from the Wood)
when I cried oomphteenth time, he responded with pretty much same sentence
that was first time when I seriously fired up LTSpice (being ignorant Greedy Boy with napkin and calculator) and then I simply couldn't confirm heavy load thesis
what I confirmed with sims , really not in a mood later to confirm in vivo with my precioussss Papa's Koan M2 , is this*** :
D1-D3 are guarding those fragile gates in case of idiotic sine driving them (load value pretty irrelevant) clamping them to similar values above positive and negative rails , also resulting in symmetrical clipping
LM385 are saving opto LED of nasty current spikes , in all big current sucking conditions , be it moderate voltage/heavy load or heavy voltage/moderate load
disclaimer - sim is sim (especially when made by Mighty ZM) , and Mithrandir is Mithrandir****
***I'm lazy to repeat and to post screenshots
****(finding great fun not just pulling someone else's leg , but even his own , here and there)

Last edited:
however,whatever explanation is , there is no single redundant part
I really have that schm as poster in my workshop , A2 size
I really have that schm as poster in my workshop , A2 size
D1-D3 are guarding those fragile gates in case of idiotic sine driving them (load value pretty irrelevant) clamping them to similar values above positive and negative rails , also resulting in symmetrical clipping
LM385 are saving opto LED of nasty current spikes , in all big current sucking conditions , be it moderate voltage/heavy load or heavy voltage/moderate load
ZM is merely playing straight man so that I can deliver punch lines...
D1 and D3 also make the amp limit symmetrically into shorts, and the
unequal Vgs between N and P require two diodes on one side.
Without 385's, the amplifier will slowly un-bias when output current
exceeds 2X bias rating. By carefully limiting single-ended bias sense,
the amplifier is still at proper bias after high current.
😎
I am going to state this in the most eloquent manner:Without 385's, the amplifier will slowly un-bias when output current
exceeds 2X bias rating. By carefully limiting single-ended bias sense,
the amplifier is still at proper bias after high current.
😎
You're a bloody genius.
🙂
Thank you sir. Maybe someday I will get a chance to build this.
If I am really fortunate I may someday understand your brilliance.
If I am really fortunate I may someday understand your brilliance.
I am going to state this in the most eloquent manner:
You're a bloody genius.
🙂
after several years of having that schematic , also having amp in my living room , I'm still flabbergasted
I remember , from dusty papers , amps similar in both simplicity and execution
however , uniqueness of this one is fact that is brought in this era , both in spite and due to modern era parts and habits of how to use them
you'll hardly find both simpler (or even "just" equally simple) and more capable of doing what is made for
True Papa's Koan , isn't it .
I am going to state this in the most eloquent manner:
You're a bloody genius.
🙂
Absolutely! Well said...
OK , I understand that Greedy Boyz are happy and thrilled .......... but - how on earth , Pa can discontinue that amp ?!!


When Pa runs out of new ideas he'll bring back special anniversary versions of the old amps with possibly a few new tweaks.OK , I understand that Greedy Boyz are happy and thrilled .......... but - how on earth , Pa can discontinue that amp ?!!
![]()
However I have a sneaky suspicion Papa is not short on ideas, so that could be quite a while away.
Hahahaha
Just because it's there doesn't mean people have to use it.Hello,
cascode version...
But I don´t know if there is a really demand?
The M2 is cheap, the costs are mainly the selected JFET´s, the transformer(s) and the enclosure.
Jean-Paul
The beauty of it is if you have jfets with idss above 10mA you can use them with cascode.
Also if more power is needed or someone has transformers with 24 secondaries then that can all be taken care of, which can't be performed without cascoding the jfets.
I mean, is there a really demand for PCB?
I try to define HIFI2000 enclosures in 3D for simulated complete M2 assembly (depends on quality of the 2D HIFI2000 documentation).
Idea for the supplies?
Jean-Paul
I try to define HIFI2000 enclosures in 3D for simulated complete M2 assembly (depends on quality of the 2D HIFI2000 documentation).
Idea for the supplies?
Jean-Paul
I mean, is there a really demand for PCB?
I try to define HIFI2000 enclosures in 3D for simulated complete M2 assembly (depends on quality of the 2D HIFI2000 documentation).
Idea for the supplies?
Jean-Paul
They already sell CRC supplies in the diyaudio store. You could do a pcb but the store pcbs are quite good.
The only other thing I can think of is maybe the option for an additional pair of output devices for M2 turbo build.
There's so many nice amps to build now. It's hard to know if people will want an M2 Turbo.
There's so many nice amps to build now. It's hard to know if people will want an M2 Turbo.
- Home
- Amplifiers
- Pass Labs
- Official M2 schematic