FWIW Dept.
I've been reminded with this assembly exercise of the value of a lead bending form. Anything of a suitable width around which to bend component leads is a real time-saver. 15-20 seconds for one resistor is not much, but over the course of 40 items it mounts up. At one time I used a piece of wood cut to shape. Yesterday I used a wooden spring loaded clothes pin ( remember clothes lines? ) as my bending form. Get one and keep it in your tool kit. It also serves as a clamp for surface mount devices when you need one. The jaws can be cut to fit. If the jaws aren't deep enough you can gain some extension with another piece of wood. (Save some frozen treat sticks)
It just goes on and on.....
I've been reminded with this assembly exercise of the value of a lead bending form. Anything of a suitable width around which to bend component leads is a real time-saver. 15-20 seconds for one resistor is not much, but over the course of 40 items it mounts up. At one time I used a piece of wood cut to shape. Yesterday I used a wooden spring loaded clothes pin ( remember clothes lines? ) as my bending form. Get one and keep it in your tool kit. It also serves as a clamp for surface mount devices when you need one. The jaws can be cut to fit. If the jaws aren't deep enough you can gain some extension with another piece of wood. (Save some frozen treat sticks)
It just goes on and on.....
a couple of questions/ideas for a next version...
Hi Mauro,
I've got a couple of questions/ideas about your very interesting design.
1)
If I remember correctly, you've done some experiments with current drive... so I wonder: would it be possible to add some amount of current feedback to your "my_ref" amplifier so to possibly make its output impedance variable from ~0 (as is, i.e. without current feedback) to say some 50 ~ 100ohm?
I mean, have you tried to do so and/or do you foresee any major problem doing something like that (stability, etc)?
(I ask because, as you may know, I have done some experiments with my toy mixed-feedback modified NIGC with adjustable output impedance -using "normal", multi-way loudspeakers- and I got some quite interesting results...).
2)
I've read about the sort of "soft clipping" trick embedded in the PSU of my_ref. That's quite interesting... actually, I do believe that the "overdrive"/"overload" (i.e., "clipping") behavior of any amplifier is paramount for its perceived quality under normal use.
BTW, I was thinking about a more direct and "drastic" (and general) solution to provide a very soft and controlled "clipping" in any overdrive and/or overload condition and prevent any related problem... (such as "blocking", etc).
Imagine adding some circuitry which monitors voltages and currents in sensible points of the amplifier circuit (e.g. in "my_ref" the outputs of the 3886s and 318s) and gradually reduces the amplitude of the input signal (via a controlled attenuator placed right at the input of the amplifier... maybe simply a FET shunting a resistor...) whenever any of the monitored parameters approaches a "critical" value (e.g. the out current approach its max value, or the output voltage of one of the chip approaches its Vmax = Vsupply - Vdropout), that is before the circuit runs out of its normal ("linear") operating conditions.
In other words, something like a sort of "AGC" which prevents the amplifier entering any possible "erratic" condition (due to overdrive/overload, that is) by automatically "turning down the volume" at the input!
(...possibly down to zero in case of a short circuit on the output!)
What do you think about that?
P.S.: if you prefer discussing this in italian, please contact me directly by e-mail...
Hi Mauro,
I've got a couple of questions/ideas about your very interesting design.
1)
If I remember correctly, you've done some experiments with current drive... so I wonder: would it be possible to add some amount of current feedback to your "my_ref" amplifier so to possibly make its output impedance variable from ~0 (as is, i.e. without current feedback) to say some 50 ~ 100ohm?
I mean, have you tried to do so and/or do you foresee any major problem doing something like that (stability, etc)?
(I ask because, as you may know, I have done some experiments with my toy mixed-feedback modified NIGC with adjustable output impedance -using "normal", multi-way loudspeakers- and I got some quite interesting results...).
2)
I've read about the sort of "soft clipping" trick embedded in the PSU of my_ref. That's quite interesting... actually, I do believe that the "overdrive"/"overload" (i.e., "clipping") behavior of any amplifier is paramount for its perceived quality under normal use.
BTW, I was thinking about a more direct and "drastic" (and general) solution to provide a very soft and controlled "clipping" in any overdrive and/or overload condition and prevent any related problem... (such as "blocking", etc).Imagine adding some circuitry which monitors voltages and currents in sensible points of the amplifier circuit (e.g. in "my_ref" the outputs of the 3886s and 318s) and gradually reduces the amplitude of the input signal (via a controlled attenuator placed right at the input of the amplifier... maybe simply a FET shunting a resistor...) whenever any of the monitored parameters approaches a "critical" value (e.g. the out current approach its max value, or the output voltage of one of the chip approaches its Vmax = Vsupply - Vdropout), that is before the circuit runs out of its normal ("linear") operating conditions.
In other words, something like a sort of "AGC" which prevents the amplifier entering any possible "erratic" condition (due to overdrive/overload, that is) by automatically "turning down the volume" at the input!
(...possibly down to zero in case of a short circuit on the output!
) What do you think about that?
P.S.: if you prefer discussing this in italian, please contact me directly by e-mail...
Ok, that's as much as I can do. Waiting on more parts now.
I've installed all the remaining caps, the transistors and bridges that I have.
Also raised the 1w resistors, flipped Conn2 as per Brian and Russ's suggestions.
My relay has a dent in one end, hope there is a bit of clearance inside...
Clothes peg sounds like a good idea, must get one of them (we don't have a clothes line, I might have to steal one from the neighbours...
)
Thanks,
Michael
I've installed all the remaining caps, the transistors and bridges that I have.
Also raised the 1w resistors, flipped Conn2 as per Brian and Russ's suggestions.
My relay has a dent in one end, hope there is a bit of clearance inside...
Clothes peg sounds like a good idea, must get one of them (we don't have a clothes line, I might have to steal one from the neighbours...
)
An externally hosted image should be here but it was not working when we last tested it.
Thanks,
Michael
Help Please
I just know I'm suffering from either screen blindness or senile dementia here BUT - I remember reading a post a few days back which I'm 99% sure was from Mauro describing the connections / uses of conn2 and conn3 - 1 was for an LED the other I forget.
Well I've been trying to find that post again without success - can anyone point me towards it please - oh and I gladly accept whatever insults are coming my way when it's pointed out that it's 1 page back or something equally silly - I have tried looking for it but I think I've missed it,
thanks
Dave
I just know I'm suffering from either screen blindness or senile dementia here BUT - I remember reading a post a few days back which I'm 99% sure was from Mauro describing the connections / uses of conn2 and conn3 - 1 was for an LED the other I forget.
Well I've been trying to find that post again without success - can anyone point me towards it please - oh and I gladly accept whatever insults are coming my way when it's pointed out that it's 1 page back or something equally silly - I have tried looking for it but I think I've missed it,
thanks
Dave
Thanks !
Wow that was quick Russ - thanks
I will do what I should have done in the first place and look at the schematics! The LED is really all I want - "power on" and this design justifies a nice blue one!
Thanks again Russ
cheers
Dave.
p.s. if I can find my camera I'll post a pic of my assembled board.
Wow that was quick Russ - thanks
I will do what I should have done in the first place and look at the schematics! The LED is really all I want - "power on" and this design justifies a nice blue one!
Thanks again Russ
cheers
Dave.
p.s. if I can find my camera I'll post a pic of my assembled board.
Hi Paolo,
You set me various matters.
To begin, say you that have done a description on the Italian forum of the MY_REF philosophy.
In connection with the Zout upsurge, if you insert a res among 2 and 6 of LM318 reduces the global gain margin ( and the DF ). The bridge has a Zout of about 500 ohms, then you have to have a gain of about 5 on LM318 ( Es: 15K among 2 and 6) . In this case the Zin becomes of 3K, and have to use an input c of 10-22uF.
The stability improves to the diminish of the LM318 gain.
Soft clipping: I have set me the problem, and in a future version would be able elaborate a suitable net. Per hour I have corresponded that the clipping is " strongly " but don't do "destructive" in the load comparisons. Affair remembers that LM3886 has a structure of very efficient protection, and LM318 not suffers particular shock. A problem of base is that the limitation nets interact much with the circuit, and they serve modify all the compensations, to times with negative results. ( are enough a few pF to jeopardize any characteristics )
My idea of base is to leave the greater " movement border " possible to LM318, even to the detritment of "limit" conditioning.
Ciao
Mauro
You set me various matters.
To begin, say you that have done a description on the Italian forum of the MY_REF philosophy.
In connection with the Zout upsurge, if you insert a res among 2 and 6 of LM318 reduces the global gain margin ( and the DF ). The bridge has a Zout of about 500 ohms, then you have to have a gain of about 5 on LM318 ( Es: 15K among 2 and 6) . In this case the Zin becomes of 3K, and have to use an input c of 10-22uF.
The stability improves to the diminish of the LM318 gain.
Soft clipping: I have set me the problem, and in a future version would be able elaborate a suitable net. Per hour I have corresponded that the clipping is " strongly " but don't do "destructive" in the load comparisons. Affair remembers that LM3886 has a structure of very efficient protection, and LM318 not suffers particular shock. A problem of base is that the limitation nets interact much with the circuit, and they serve modify all the compensations, to times with negative results. ( are enough a few pF to jeopardize any characteristics )
My idea of base is to leave the greater " movement border " possible to LM318, even to the detritment of "limit" conditioning.
Ciao
Mauro
Hi Riccardo,
This is for the Vishay resistors that were selected for the parts group buy. I'm sure it applies to any brand, it makes sense when you look at it.
You can find them on the Mouser website here
If you download the datasheet, look on page 3 for the mounting diagram.
Regards,
Michael
riccardo said:
This is for the Vishay resistors that were selected for the parts group buy. I'm sure it applies to any brand, it makes sense when you look at it.
You can find them on the Mouser website here
If you download the datasheet, look on page 3 for the mounting diagram.
Regards,
Michael
Hello Michael.
I've seen the vishay's resistor datasheet.
but the distance from pcb don't depend by the voltage through the resistor?
these resistors are for voltage up 450!.
i don't know if this could be the case, but the resistor , when current flow go through, they vibrates....and could introduce noise in the circuit.
......
??
Ciao
RR
I've seen the vishay's resistor datasheet.
but the distance from pcb don't depend by the voltage through the resistor?
these resistors are for voltage up 450!.
i don't know if this could be the case, but the resistor , when current flow go through, they vibrates....and could introduce noise in the circuit.
......
??
Ciao
RR
Riccardo,
Maybe, I really don't know. Vishay only shows 1 mounting position: 8mm +/- 2mm, so you can mount them as low as 6mm and be in spec.
Wouldn't current have a bigger heating effect than voltage? Brian and Russ reckon they do get hot...
What I do know, is that if you mount them high, you can always shorten them.
But if you mount them low, and it's getting too hot, you will have to replace them to get them higher because you will cut the leads after you solder them in.
I don't know about this vibrating effecting the sound thing. But you can experiment with it if you start high
Maybe, I really don't know. Vishay only shows 1 mounting position: 8mm +/- 2mm, so you can mount them as low as 6mm and be in spec.
Wouldn't current have a bigger heating effect than voltage? Brian and Russ reckon they do get hot...
What I do know, is that if you mount them high, you can always shorten them.
But if you mount them low, and it's getting too hot, you will have to replace them to get them higher because you will cut the leads after you solder them in.
I don't know about this vibrating effecting the sound thing. But you can experiment with it if you start high
Hi Mauro,
mmmh... searching... it is this one, right?
http://www.audiocostruzioni.com/forum/topic.asp?ARCHIVE=&whichpage=2&TOPIC_ID=15377
so it should be doable without much troubles... great!
given that it so simple... could you do this test? Try to use a resistor value as to get a Zout (closed loop) of about 8 ~ 16 ohm... listen to the amplifier and let us know your listening impressions!
(I would bet on the results, but will not anticipate them to not bias you in case you'll do it. )
I'm planning to build and evaluate your circuit as is, then try the mod myself... but that will take some time and I'm way too curious to wait that much for the results!
BTW: about the Zin: that's the only serious concern I have. I need some ~50K to avoid degrading the performance of the tube output stage of my CDP (it's made with an ECC82 connected in common cathode mode, without any feedback).
Do you think is it possible to increase the Zin of your amplifier up to about 40K ~ 50K without compromising performances?
of course... the short circuit/overload protection is just a (though welcome) side effect; the intended pourpose of the idea was to make sure that -under normal usage condition- the circuit will never be pushed out of its limits. The idea behind it is an old one, first suggested (AFAIK) by N.H.Crowhurst a loong time ago... (I guess you know it well!).
That is, during normal use, there may be a lot of different events (either "external", like e.g. defects on the disk, EMI/RFI, spikes on the power line, etc, or in the program itself) which may create short but possibly very strong transients.
Those are normally way too short to be directly audible, but they may well push the amplifier out of its limits... this in turn will usually produce some much longer transients until the circuit recover to its normal state. And these effect are definitely well audible, though not directly but as a general degradation of the perceived quality.
mmmh... don't know if we are speaking about the same thing! My idea was of using an independent "servo" circuit using a few extra op-amps (likely 4 per channel) to do the job, so to minimize the interactions with the main circuit. The "sensors" should have very high input impedance and would be connected to pin 6 of the LM318s, pin 3 of the 3886s and the outputs.
I was supposing that such high impedances should not do any harm in those points of the circuit... or am I wrong, and even there the circuit is so sensitive that the added capacitances due to connection of "sensors" input may mess things up?
About the servo "action", I would do it trough a voltage controlled attenuator (e.g. an "L" or "T" net with a -normally non conducting!- FET on the "vertical" leg, but possibly something much better... which then maybe can be used also for the normal volume control), connected either before or right after the input decoupling capacitor C13/C29.
There I guess it should do no harm either... otherwise the circuit would be way too much sensible to the output impedance of the source (or preamplifier), interconnects cable capacitance, etc...
sure... understood (and mostly agreed).
mauropenasa said:
mmmh... searching... it is this one, right?
http://www.audiocostruzioni.com/forum/topic.asp?ARCHIVE=&whichpage=2&TOPIC_ID=15377
so it should be doable without much troubles... great!
given that it so simple... could you do this test? Try to use a resistor value as to get a Zout (closed loop) of about 8 ~ 16 ohm... listen to the amplifier and let us know your listening impressions!
(I would bet on the results, but will not anticipate them to not bias you in case you'll do it.
)I'm planning to build and evaluate your circuit as is, then try the mod myself... but that will take some time and I'm way too curious to wait that much for the results!
BTW: about the Zin: that's the only serious concern I have. I need some ~50K to avoid degrading the performance of the tube output stage of my CDP (it's made with an ECC82 connected in common cathode mode, without any feedback).
Do you think is it possible to increase the Zin of your amplifier up to about 40K ~ 50K without compromising performances?
of course... the short circuit/overload protection is just a (though welcome) side effect; the intended pourpose of the idea was to make sure that -under normal usage condition- the circuit will never be pushed out of its limits. The idea behind it is an old one, first suggested (AFAIK) by N.H.Crowhurst a loong time ago... (I guess you know it well!).
That is, during normal use, there may be a lot of different events (either "external", like e.g. defects on the disk, EMI/RFI, spikes on the power line, etc, or in the program itself) which may create short but possibly very strong transients.
Those are normally way too short to be directly audible, but they may well push the amplifier out of its limits... this in turn will usually produce some much longer transients until the circuit recover to its normal state. And these effect are definitely well audible, though not directly but as a general degradation of the perceived quality.
mmmh... don't know if we are speaking about the same thing!
My idea was of using an independent "servo" circuit using a few extra op-amps (likely 4 per channel) to do the job, so to minimize the interactions with the main circuit. The "sensors" should have very high input impedance and would be connected to pin 6 of the LM318s, pin 3 of the 3886s and the outputs. I was supposing that such high impedances should not do any harm in those points of the circuit... or am I wrong, and even there the circuit is so sensitive that the added capacitances due to connection of "sensors" input may mess things up?
About the servo "action", I would do it trough a voltage controlled attenuator (e.g. an "L" or "T" net with a -normally non conducting!- FET on the "vertical" leg, but possibly something much better... which then maybe can be used also for the normal volume control), connected either before or right after the input decoupling capacitor C13/C29.
There I guess it should do no harm either... otherwise the circuit would be way too much sensible to the output impedance of the source (or preamplifier), interconnects cable capacitance, etc...
sure... understood (and mostly agreed).
Any considerations:
By Paulo:
If you want use the current pump autonomously ( for the current tests ) are able use the only bridge, with a input buffer ( to maintain balance the resistances of the bridge ) .
If the output is closed on a load, the circuit is stable. If you read the Thread that I have done finds the parameters of sizing of the bridge. For the use in current enough it remove the 100 ohm res. after among in and out...
http://www.diyaudio.com/forums/showthread.php?threadid=55925&highlight=
All:
"Hot" Res.: if you leave a space between the PCB and the better reses the thermal ventilation, and the superficial temperature of the component diminishes. This method is ordinary, in the circuits to tall dissipation. It is good rule use this technique even by R3 and R26, even if in the use normal doesn't heat much...
My_Ref updating: I have not the time of behaviour other changes in the next few months. My idea is to pick up the general impressions of the members, and eventually think to a "plus" version. As the characteristics of this circuit am enough "unique" creed that will leave the project to this levels, and eventually I will think about something that not it replaces but with characteristic diverged ( example more power, etc.). I consider the RevA ( and Rev3 ) as complete level of develop hardware ( even because is not an audio planner!).
In connection with the "better proposals": How to have explained, My_Ref is a "my own " amp, for which is implicit that the choice of the components and of circuit it corresponds to my "thought". I don't succeed to find un "weak spots" in the sound of this circuit. When it finds them resolve modifying.Don't ask me if "sound better" changing the cap or the res: for me better than don't maids, because it is already better than very "great" amps!
Ciao
Mauro
By Paulo:
If you want use the current pump autonomously ( for the current tests ) are able use the only bridge, with a input buffer ( to maintain balance the resistances of the bridge ) .
If the output is closed on a load, the circuit is stable. If you read the Thread that I have done finds the parameters of sizing of the bridge. For the use in current enough it remove the 100 ohm res. after among in and out...
http://www.diyaudio.com/forums/showthread.php?threadid=55925&highlight=
All:
"Hot" Res.: if you leave a space between the PCB and the better reses the thermal ventilation, and the superficial temperature of the component diminishes. This method is ordinary, in the circuits to tall dissipation. It is good rule use this technique even by R3 and R26, even if in the use normal doesn't heat much...
My_Ref updating: I have not the time of behaviour other changes in the next few months. My idea is to pick up the general impressions of the members, and eventually think to a "plus" version. As the characteristics of this circuit am enough "unique" creed that will leave the project to this levels, and eventually I will think about something that not it replaces but with characteristic diverged ( example more power, etc.). I consider the RevA ( and Rev3 ) as complete level of develop hardware ( even because is not an audio planner!).
In connection with the "better proposals": How to have explained, My_Ref is a "my own " amp, for which is implicit that the choice of the components and of circuit it corresponds to my "thought". I don't succeed to find un "weak spots" in the sound of this circuit. When it finds them resolve modifying.Don't ask me if "sound better" changing the cap or the res: for me better than don't maids, because it is already better than very "great" amps!
Ciao
Mauro