ostripper said:
Yes. If QX is the same type of transistor as Q8 and the 4.7k resistor is reduced in value so that the emitter current of QX is the same as that of Q8 then the base current pulled from each leg of the VAS will be equal.
However if you used an MJE350 for QX and forced that much current through it you'd hardly have an improved circuit.
You can buffer the VAS with an EF (ala D.Self) while using an identical device for QX. If the emitter resistors are selected so that the emitter currents of these transistors are equal, then you get the same base current drawn from each leg of the LTP.
Cheers,
Glen
Edit:
See attached pic. To simplify I am ignoring the current mirror transistor and VAS transistor base currents as they are negligible and I have assumed Vbe=0.7V for all transistors, but you should get the idea.
Attachments
AKSA said:
Textbook is the word.
(in swedish we say: skolboksexempel =schoool-book-example)
I have tried many times and think I make easily read schematics.
Now making a 'perfect schema' to pubish is not made on a coffeebreak.
More like it hour(s) of work!!!
But many times, I know from my own experience, that I do not even bother to try to understand some circuits,
and even less probably dicuss and comment,
if I have to spend more than like 15 minutes to try to see what is happening in the circuit.
And this is a pity.
Some circuit may be great .. but the presentation does not match the circuitry.
A bit like in marketing:
The package, advertising, colours and the looks is half the product
I understand fully ,glen, I will make 2 more boards w/these
changes. Now I am going to just use the "base design"
to ensure reliability.
Not quite in a american coffee break... but close, I used eagle
for schema, imported into paintshop, used text tool,all done.
Wait till you see the protection circuit, 5 IC's- 12 diodes - 2
transistors...
OS
changes. Now I am going to just use the "base design"
to ensure reliability.
Not quite in a american coffee break... but close, I used eagle
for schema, imported into paintshop, used text tool,all done.
Wait till you see the protection circuit, 5 IC's- 12 diodes - 2
transistors...
OS
I would prefer the box ring with him...but i have to agree
Kleinschmidt is rigth about the comparison method... we humans.... we are ready to fool ourselves all time long.... comparison testings needs some method.... you cannot even know the choice is playing... cannot know what the switch is doing.
This way evaluation will be only listening....no passion, no bias, no previously decided result that we all have, and always we have preferences....other may flip switch.... instantaneous switching... cannot have delay.... audio memory not good to comparison...better to listen A and B almost simultaneously.... time given only to inform:
- Now playing A
- Now playing B
2 seconds maximum waiting the one is listening to understand words and memorize that A or B is playing... and more 3 seconds listening..not allowing time to brain proceed natural adjustment....unfortunattelly i found that five cicles..or 25 seconds is enougth to brain adjust both...brain passes to perceive, for instance, less agressive treble..less harshing.. and do that into both.... have to interrupt and change type of musical program.
If comparison is about two amplifiers.... we cannot see them.... also cannot know brands (some are famous and will have strong influences...positive or negative depending the evaluator character).... when amplifier is pretty, half way gone to have the better evaluation.... we are not microphones... reason why people prefer microphones and instruments to evaluate...we are complex, we have intelligence...we go capturing everything....including the guy that is switching face expressions will be observed... voice modulation... how large is your eye Iris... we have many informations..not only sound pressure into ears... have to remove all other stuff not to polute testings.
I would be happy if i could be a good guy to make testings.... i have studied a lot, and tried a lot..but as i am everyday learning something new, i feel i am thousand miles distant from the perfection..... BUT... i see that a lot of folks are thousand miles behind me into this subject too.
I think it is very difficult the procedure...need experienced, skilled folks controlling...yes... psychologist is a very good idea, as they use to understand the trick and traps we use to play with ourselves.
So... A plays 5 seconds and also B plays 5 seconds.... better the one that is switch operating also be ignorant, ignore what is going on...not to modulate voice with his preference...this is captured too...people is not idiot... we humans, we sniff things alike dogs.
The ring to box is a very good idea...i have already two to invite to a nice fun with big fat Charlie.
regards,
Carlos
Kleinschmidt is rigth about the comparison method... we humans.... we are ready to fool ourselves all time long.... comparison testings needs some method.... you cannot even know the choice is playing... cannot know what the switch is doing.
This way evaluation will be only listening....no passion, no bias, no previously decided result that we all have, and always we have preferences....other may flip switch.... instantaneous switching... cannot have delay.... audio memory not good to comparison...better to listen A and B almost simultaneously.... time given only to inform:
- Now playing A
- Now playing B
2 seconds maximum waiting the one is listening to understand words and memorize that A or B is playing... and more 3 seconds listening..not allowing time to brain proceed natural adjustment....unfortunattelly i found that five cicles..or 25 seconds is enougth to brain adjust both...brain passes to perceive, for instance, less agressive treble..less harshing.. and do that into both.... have to interrupt and change type of musical program.
If comparison is about two amplifiers.... we cannot see them.... also cannot know brands (some are famous and will have strong influences...positive or negative depending the evaluator character).... when amplifier is pretty, half way gone to have the better evaluation.... we are not microphones... reason why people prefer microphones and instruments to evaluate...we are complex, we have intelligence...we go capturing everything....including the guy that is switching face expressions will be observed... voice modulation... how large is your eye Iris... we have many informations..not only sound pressure into ears... have to remove all other stuff not to polute testings.
I would be happy if i could be a good guy to make testings.... i have studied a lot, and tried a lot..but as i am everyday learning something new, i feel i am thousand miles distant from the perfection..... BUT... i see that a lot of folks are thousand miles behind me into this subject too.
I think it is very difficult the procedure...need experienced, skilled folks controlling...yes... psychologist is a very good idea, as they use to understand the trick and traps we use to play with ourselves.
So... A plays 5 seconds and also B plays 5 seconds.... better the one that is switch operating also be ignorant, ignore what is going on...not to modulate voice with his preference...this is captured too...people is not idiot... we humans, we sniff things alike dogs.
The ring to box is a very good idea...i have already two to invite to a nice fun with big fat Charlie.
regards,
Carlos
Re: I would prefer the box ring with him...but i have to agree
Yes, comparison testing can be very difficult if the differences are small. You are right Carlos in that the memory is short. There is also the fatigue factor, thus A/B/A/B/A. is more than enough to confirm any real listenable difference, and the listens don't have to be all that long as the differences are either there or they are not.
But, as long as you are honest in your listening, you don't need all the hallubaloo with blind testing. Why would you want to fool yourself if you are the one seeking an honest answer? But it does depend on WHAT you are comparing and WHO is comparing!
destroyer X said:
Yes, comparison testing can be very difficult if the differences are small. You are right Carlos in that the memory is short. There is also the fatigue factor, thus A/B/A/B/A. is more than enough to confirm any real listenable difference, and the listens don't have to be all that long as the differences are either there or they are not.
But, as long as you are honest in your listening, you don't need all the hallubaloo with blind testing. Why would you want to fool yourself if you are the one seeking an honest answer? But it does depend on WHAT you are comparing and WHO is comparing!
When we are testing other guys ideas we can be reasonably fair and honest
And this depends your relationship with the guy... empathy.
But if you are evaluating your own ideas, against other ideas that does not belongs to you.... for sure you would prefer your ideas to be the winner...so... you will not be a fair evaluator.
My dear friend...will tell you something you may not know.
We use to lye to ourselves all time long.... when you buy a gift and give it to someone, you may believe you are doing that to make this other guy, "the someone" happy..... maybe a little..but the real thing is that:
- You will feel great, you gonna feel good, you gonna feel satisfied, you gonna feel happy...
- The guy that received a gift from you will be a nice friend, will not be dangerous anymore, an allied... and will have to pay this gift somewere or someway..for sure he will have to pay that.
- Also you are persuading him to accept you as a friend, you will be buying his attention and tollerance against your own personal character defects...
Man... this can produce a book... and this is only an example.... fair?....no one is entirelly fair to judge.... because of that exist laws and lawyers and people to judge.... a lot of carefull to reduce risks...and even this way a lot of unfair things happens.
I would say you have lied to yourself, today, at least 100 times...even when you think you lye with yourself.... we have some ideal of ourselves..a constructed human that is not the real one..and you negotiate with yourself all time long to believe you are that ideal one...well..... complicated to explain to people that had no psychologic trainment.
regards,
Carlos
And this depends your relationship with the guy... empathy.
But if you are evaluating your own ideas, against other ideas that does not belongs to you.... for sure you would prefer your ideas to be the winner...so... you will not be a fair evaluator.
My dear friend...will tell you something you may not know.
We use to lye to ourselves all time long.... when you buy a gift and give it to someone, you may believe you are doing that to make this other guy, "the someone" happy..... maybe a little..but the real thing is that:
- You will feel great, you gonna feel good, you gonna feel satisfied, you gonna feel happy...
- The guy that received a gift from you will be a nice friend, will not be dangerous anymore, an allied... and will have to pay this gift somewere or someway..for sure he will have to pay that.
- Also you are persuading him to accept you as a friend, you will be buying his attention and tollerance against your own personal character defects...
Man... this can produce a book... and this is only an example.... fair?....no one is entirelly fair to judge.... because of that exist laws and lawyers and people to judge.... a lot of carefull to reduce risks...and even this way a lot of unfair things happens.
I would say you have lied to yourself, today, at least 100 times...even when you think you lye with yourself.... we have some ideal of ourselves..a constructed human that is not the real one..and you negotiate with yourself all time long to believe you are that ideal one...well..... complicated to explain to people that had no psychologic trainment.
regards,
Carlos
This always seems to happen.
Some offers up a frugal amp and before you know it everyone has chipped in and it has grown into a monster.
I have more frugal designs than this amp by far and have no problems with sound quality.
Using 3 transistors in a current mirror is just silly.
My 2 transistor CCS work just fine.
In fact a single transistor CCS would probably do the job.
Some offers up a frugal amp and before you know it everyone has chipped in and it has grown into a monster.
I have more frugal designs than this amp by far and have no problems with sound quality.
Using 3 transistors in a current mirror is just silly.
My 2 transistor CCS work just fine.
In fact a single transistor CCS would probably do the job.
Do I note a bit o' sarcasm..
I'm still keeping it frugal ,BC-xxx's are $.04 apiece and you can
cram alot of them on the input section of a board.
I've been doing high power testing and have a few minor bugs
1. changed C1 input cap to 2.2 uf bipolar (are el caps bad?)
better bass. Got rid of RC phase comp. across CM collectors
(why add another pole?)
2. I start the bias at 11mV across emitter resistor (50ma),
as amp heats up it goes to about 15mV, then when it
really gets warm it levels out at 17mV and sometimes
backs off a little.
Shut amp off, 11mV again after cooling. Does this sound
normal??
3. How do you get more gain from an amp ? Almost have to
overdrive my soundcard to get descent output.
Positive factors..
NO turn on thump, at all.
Very detailed highs, am hearing fingers strumming strings,
breath, and highs are clearly hitting SPL's (tickle the eardrums)
NO noise (ear right on tweeter).
Drove 20' speaker cable run at high volumes without zoble
(rear speakers)no instability or signs of oscillation.
Output offset stays + - 2mv ,all conditions.
Here it is in it's present form...with readings..
To be objective , I first want to perfect the base design before I
"pop out" another set of boards.
OS
I'm still keeping it frugal ,BC-xxx's are $.04 apiece and you can
cram alot of them on the input section of a board.
I've been doing high power testing and have a few minor bugs
1. changed C1 input cap to 2.2 uf bipolar (are el caps bad?)
better bass. Got rid of RC phase comp. across CM collectors
(why add another pole?)
2. I start the bias at 11mV across emitter resistor (50ma),
as amp heats up it goes to about 15mV, then when it
really gets warm it levels out at 17mV and sometimes
backs off a little.
Shut amp off, 11mV again after cooling. Does this sound
normal??
3. How do you get more gain from an amp ? Almost have to
overdrive my soundcard to get descent output.
Positive factors..
NO turn on thump, at all.
Very detailed highs, am hearing fingers strumming strings,
breath, and highs are clearly hitting SPL's (tickle the eardrums)
NO noise (ear right on tweeter).
Drove 20' speaker cable run at high volumes without zoble
(rear speakers)no instability or signs of oscillation.
Output offset stays + - 2mv ,all conditions.
Here it is in it's present form...with readings..
To be objective , I first want to perfect the base design before I
"pop out" another set of boards.
OS
Please guys,
I'm not going to throw any dirt or name names, and will readily admit that most of what has been said on here about improving his amp has been said by guys way smarter than I am... BUT remember this guys goals.
He is starting out. Ive been there. He's designed a pretty good circuit given this, and that his goal was to keep it simple and cheap. There's been a massive argument here over the use of a single transistor. It's not exactly encouraging.
ostripper - just in case my posts got lost in the mayhem, I would make a few suggestions:
1) Add an extra pair of outputs, or cut your rails down to 48V. Two pairs of outputs on 58V is pushing it.
2) Buffer the VAS. As you seem to be reading Doug Self's work, you will see what I mean on his "Blameless" amplifier. A BC556 will work well.
3) Instead of MJE340/350 look for 2SA1381/2SC3503. Fairchild second source these (using KSA/KSC prefixes) and you can get some direct from them as samples for a very small cost.
They are excellent VAS devices and it is worth the small cost to use them over the MJE340/350. They should drop right in although you may need to tweak your compensation a little.
I'm not going to throw any dirt or name names, and will readily admit that most of what has been said on here about improving his amp has been said by guys way smarter than I am... BUT remember this guys goals.
He is starting out. Ive been there. He's designed a pretty good circuit given this, and that his goal was to keep it simple and cheap. There's been a massive argument here over the use of a single transistor. It's not exactly encouraging.
ostripper - just in case my posts got lost in the mayhem, I would make a few suggestions:
1) Add an extra pair of outputs, or cut your rails down to 48V. Two pairs of outputs on 58V is pushing it.
2) Buffer the VAS. As you seem to be reading Doug Self's work, you will see what I mean on his "Blameless" amplifier. A BC556 will work well.
3) Instead of MJE340/350 look for 2SA1381/2SC3503. Fairchild second source these (using KSA/KSC prefixes) and you can get some direct from them as samples for a very small cost.
They are excellent VAS devices and it is worth the small cost to use them over the MJE340/350. They should drop right in although you may need to tweak your compensation a little.
Frugalamp
Ostripper
Ignore the sniping, you are doing good.
Bias can be a bit touchy setting up. The main thing is it levels out again O.K. Could pay to connect your DMM across an emitter resistor with thin leads, and close the case up, while watching carefully that it stabilises O.K. when it is enclosed.
Some designs use electros of 10 x the value they would for film type input capacitors. Polypropylene sound very good, but are bulky and expensive. No input cap is better, but much harder to implement, unless your soundcard has an output coupling capacitor? Check for no D.C. out of your soundcard
Your gain is already fairly high at 34. (1K + 33K/1K)
The Windows volume control isn't set too low is it ?
+-2mV offset is pretty damn good !
Stick with it !
SandyK
P.S. All good advice from jaycee too.
Alex
Ostripper
Ignore the sniping, you are doing good.
Bias can be a bit touchy setting up. The main thing is it levels out again O.K. Could pay to connect your DMM across an emitter resistor with thin leads, and close the case up, while watching carefully that it stabilises O.K. when it is enclosed.
Some designs use electros of 10 x the value they would for film type input capacitors. Polypropylene sound very good, but are bulky and expensive. No input cap is better, but much harder to implement, unless your soundcard has an output coupling capacitor? Check for no D.C. out of your soundcard
Your gain is already fairly high at 34. (1K + 33K/1K)
The Windows volume control isn't set too low is it ?
+-2mV offset is pretty damn good !
Stick with it !
SandyK
P.S. All good advice from jaycee too.
Alex
Good idea OS to get rid of that RC. Looking into the crystal ball of the simulator I spyed that was a source of instability. 
If you don't like an EF on the VAS, changing the VAS to the 2SA1381 (as jaycee suggests, and me previously) will make a large improvement. This is worth doing.
You might want to run your VAS a tad cooler. It is running hot now for no good reason. Certainly if you do go with a less rugged device for the VAS you'll need to cut the current or kiss it goodbye.
If you don't like an EF on the VAS, changing the VAS to the 2SA1381 (as jaycee suggests, and me previously) will make a large improvement. This is worth doing.
You might want to run your VAS a tad cooler. It is running hot now for no good reason. Certainly if you do go with a less rugged device for the VAS you'll need to cut the current or kiss it goodbye.
One thing you can try later, if you wish,
is to add input bias current with compensation feature.
The bias current for input, like 7uA maybe, is now flowing via R2 and R12 = 33 kOhm.
Voltage drop across those R is maybe like 200-230 mV.
And there is some current noise from this little flow.
Thanks to C5 100uF, we have equal DC-Resistance on both sides.
Otherwise would be 33 kOhm vs (1k//33k) 970.6 Ohm,
in which case the 7uA bias would cause an enormous Offset at output.
Something like (the gain) 33 x 0.2 Volt = 6.6 Volt
If we have another source for input bias current,
we can have any values for RESISTORS,
have no idle current noise in those resistors
and we can remove C5 DC-blocking capacitor.
We can add bias current addition via one extra PNP pair.
Actually in an ideal four transistor complementary LTP,
the pnp pair and npn pair bias current will be equal and so zero bias via Resistors.
Now, in reality, the PNP pair will have different gain than NPN
and we endup with an offset current even in those 4 transistor input stages.
Evenso, the two NPN can have slightly different hfe.
So we may need to be able to adjust each transistors input current
if we want to cancel out all input bias currents.
I attach my latest spiced amplifier I am working on.
In my case, the PNP pair has got considerably higher HFE.
So I need to add some more base current for the NPN pair.
And here is my old topic with schematics how we arrange a nice
input bias current cancelling:
Amplifiers >Solid State >Offset Correction using Bias Compensation
http://www.diyaudio.com/forums/showthread.php?threadid=71055
Lineup
is to add input bias current with compensation feature.
The bias current for input, like 7uA maybe, is now flowing via R2 and R12 = 33 kOhm.
Voltage drop across those R is maybe like 200-230 mV.
And there is some current noise from this little flow.
Thanks to C5 100uF, we have equal DC-Resistance on both sides.
Otherwise would be 33 kOhm vs (1k//33k) 970.6 Ohm,
in which case the 7uA bias would cause an enormous Offset at output.
Something like (the gain) 33 x 0.2 Volt = 6.6 Volt
If we have another source for input bias current,
we can have any values for RESISTORS,
have no idle current noise in those resistors
and we can remove C5 DC-blocking capacitor.
We can add bias current addition via one extra PNP pair.
Actually in an ideal four transistor complementary LTP,
the pnp pair and npn pair bias current will be equal and so zero bias via Resistors.
Now, in reality, the PNP pair will have different gain than NPN
and we endup with an offset current even in those 4 transistor input stages.
Evenso, the two NPN can have slightly different hfe.
So we may need to be able to adjust each transistors input current
if we want to cancel out all input bias currents.
I attach my latest spiced amplifier I am working on.
In my case, the PNP pair has got considerably higher HFE.
So I need to add some more base current for the NPN pair.
And here is my old topic with schematics how we arrange a nice
input bias current cancelling:
Amplifiers >Solid State >Offset Correction using Bias Compensation
http://www.diyaudio.com/forums/showthread.php?threadid=71055
Lineup
Attachments
ostripper said:
1) Electro caps are definately evil in GEB land. That RC made no sense.
2) Yes that sounds normal, but 11mV isn't much
3) Increase the value of the feedback resistor connected between the output and the inverting input.
jaycee said:
Well the thread starter decided to design in that inconsequential but controversial third current mirror transistor himself and seems to be intent on keeping it for some reason
One has to fend off an awful lot of :bs: just to explain how to implement it properly (which is a seperate issue to whether it is worthwhile in such a design or not).
Cheers,
Glen
Glad to see you here, carlos... I acknowledge to using your
"hands on" comparison method. Today, I found this amp
has really great highs/detail but was a little weak in bass.
My daughters PC (with same soundcard) has 2 little 50w DX's
with same speakers and it had better bass. Not as loud, but
better bass.DX has 4.7uf EL cap so I added added 2.2uf to
new amp .. better bass, but with the highs. Other tricks
that work (and can be heard) are VAS current ,CM balance..etc.
I might even pull the MJE340 (Q6) and try a bootstrap current
source.
OS
Re: When we are testing other guys ideas we can be reasonably fair and honest
Carlos,
This is all a side post and off topic, however, you are making general comments about comparing almost anything and lying and fooling yourself. I am not. Big difference, but we let it be.
I always like your posts as they are such a wonderful thing with the language and it's use in expression.
Now back to reading the struggles in this thread.....
Steve
destroyer X said:
Carlos,
This is all a side post and off topic, however, you are making general comments about comparing almost anything and lying and fooling yourself. I am not. Big difference, but we let it be.
I always like your posts as they are such a wonderful thing with the language and it's use in expression.
Now back to reading the struggles in this thread.....
Steve
that's 50ma
across 1 emitter resistor would you suggest more.? at a
cold start.
right now watching movies it is 14mv - 65-70 ma.
As far as the buffered current mirror I might go the route
of a balanced diff. VAS which would eliminate the problem
entirely or go the self way and just EF the existing VAS to
at least minimize the issue. This all started with 1 little 2 cent
diode...
Should I just build a little preamp to up the gain (sound card is
low <1V)
ostripper
I see you have totally missed my Post #74
Such things can happen.
Make sure you read it. Even if you do not have to comment.
http://www.diyaudio.com/forums/showthread.php?postid=1663180#post1663180
Lineup
I see you have totally missed my Post #74
Such things can happen.
Make sure you read it. Even if you do not have to comment.
http://www.diyaudio.com/forums/showthread.php?postid=1663180#post1663180
Lineup
Sorry lineup, I see it now ,you must of posted between my
posts, page did not update.
I see the advantages to this,active DC biasing instead of
passive. My only issue with this is:
What happens if it fails, DC ..bang,added trannies (but they are
frugal - $.04)
looks interesting, will plug it in to this amp (LT spice)and
add pads (or daughter card) to next boards (soon).
Thanks again for models,
OS
posts, page did not update.
I see the advantages to this,active DC biasing instead of
passive. My only issue with this is:
What happens if it fails, DC ..bang,added trannies (but they are
frugal - $.04)
looks interesting, will plug it in to this amp (LT spice)and
add pads (or daughter card) to next boards (soon).
Thanks again for models,
OS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.