nigelwright7557 said:
I know few things about fets, thus i accept your reference as real because i consider you honest man. But we discuss with Glen about bipolars. And in such type arrangements with a single VAS stage usually a value of standing current of 20mA for VAS it is a must so as we have similar fall time with the rise time in a square wave above 10KHz.
Fotios
G.Kleinschmidt said:
From the side of frequency response, if we consider that an amplifier has a Gaussian response (which is not far from reality) then if we apply the given equation: -3dB Bandwidth(Hz) = 0,34 / Rise time(sec), then from the fig.5 which is the worst case of a 20KHz square under a load of 8Ù the measured rise time of 1,95ìsec it gives: -3dB BW(Hz) = 0,34 / 1,95.10^-6 = 174,36KHz without TIM distortion and the Cdom fully compensated. That frequency response is not enough?
Finally which is the value of slew rate estimation? The only that i know is its relation with amplifiers mainly because can be expressed in any frequency. It is related with the tranconductance of LTP and that the open loop gain it is large if the transconductance of LTP is high, and the Cdom has a reversed relation with slew rate. Slew rate is a parameter related exclusively with amplifiers.
Instead Rise Time is an estimation parameter used in whole electronics.
I think also that, another one parameter named impulse response that this neglected although it gives very useful informations for the behaviour of any amplification arrangement.
Fotios
G.Kleinschmidt said:
Hi Glen
After a new examination of my project according to your suggestions and to your calculations, as well as those of D. Self, I have to refer that so you as D. Self (moreover it is obvious that your propositions are based on D. Self book) you are absolutelly wrong.
My project maybe it is something simillar with D. Self but in reality it operates under a different manner. Also there are some big differences such us the seperate CCS for LTP and VAS as well that the VAS stage it is thermally compensated. Maybe because these, they are not applicable the calculations and the mathematic relations that you propose on my project. In these included the increase of LTP current and the increase of the value of degeneration resistors to obtain a higher slew rate. One more thing it is that the EF coupled transistor before the VAS transistor in practice proved uselless. Either it exists or not this EF transistor, the result is the same. Therefore its existence in my project it is only for impression and nothing else.
I have one question for you: For a slew rate of 50V/ìs how much must be the rise time? Can you please make a measurement of one of your projects and you post the resulted curve here? As i do here me, the stupid, to expose in criticism myself? Ah?
We are satisfied from words, we have the need to see real works!
Well, after the examination i found how i can to reduce the rise time down to 1ìsec without decompensation of Cdom and reducing the stability. It is so simple, as the only needed it is the reduction of the value of one passive part or the increase of the value of another one passive part. Or to remove one passive part from the circuit. I quote again the schematic of my project to facilitate you. Can you find in schematic which are these two parts? I believe that for a person with a level of knoweledge such as your it is very easy.
An externally hosted image should be here but it was not working when we last tested it.
P.S.
Folks, don't trust so much the words the schematics the theories and the papers. One thing a project on the paper with schematics and mathematic calculations and other thing this project implemented in practice in the real world. You must be in position to draw a correct PCB also for example.
Fotios
Hey Everybody!
I lost the motivation to go on with the project as it took lot more time than planned, so I took a rest for the summer. But now it's time to finish the project with regained strength!🙂
I have reconsidered the schematics in a few details and would like to know, what others think of it. Then if that's ok, I will draw all the boards from scratch. So, the current version:
Main Amp
Schematics
Changes:
*Separated DC protection to another board.
*Bias now with two transistors to match characteristics of both legs of the output.
*Input current source and VAS cascode transistors get their bias through red led voltage source. This should be quite stiff reference. Driving current ~7mA.
*Quiecent currents of all the stages:
-------Outputs ~60mA(ca. 20 mV on emitter resistors)
-------Drivers ~30mA
-------Predrivers ~16mA
-------VAS ~10mA
-------Inputs ~7mA from current source
Questions:
*Is the LED biasing ok?
*Are the currents ok?
*Do I need one ground for just the input and feedback and other for supply caps of the input and VAS supply or I can unite them?
Dual Protection
This is the new board, used stereo schematics for universal board, will solder bridges for mono mode (to use both sides of relay)
Schematics
Power
Schematics
Changes:
*Bleeder for second main supply
*Corrected reference for DC protection supply
Preamp
Schematics
Soft Start
Schematics
With best wishes,
Aq
I lost the motivation to go on with the project as it took lot more time than planned, so I took a rest for the summer. But now it's time to finish the project with regained strength!🙂
I have reconsidered the schematics in a few details and would like to know, what others think of it. Then if that's ok, I will draw all the boards from scratch. So, the current version:
Main Amp
Schematics
Changes:
*Separated DC protection to another board.
*Bias now with two transistors to match characteristics of both legs of the output.
*Input current source and VAS cascode transistors get their bias through red led voltage source. This should be quite stiff reference. Driving current ~7mA.
*Quiecent currents of all the stages:
-------Outputs ~60mA(ca. 20 mV on emitter resistors)
-------Drivers ~30mA
-------Predrivers ~16mA
-------VAS ~10mA
-------Inputs ~7mA from current source
Questions:
*Is the LED biasing ok?
*Are the currents ok?
*Do I need one ground for just the input and feedback and other for supply caps of the input and VAS supply or I can unite them?
Dual Protection
This is the new board, used stereo schematics for universal board, will solder bridges for mono mode (to use both sides of relay)
Schematics
Power
Schematics
Changes:
*Bleeder for second main supply
*Corrected reference for DC protection supply
Preamp
Schematics
Soft Start
Schematics
With best wishes,
Aq
fotios said:
OK, so like D.Self I am absolutely wrong. So your circuit uses a seperate CCS for the LTP and VAS. No sure I've seen any that use a combined CCS. The rest of your post doesn't make a lot of sense either.
Your questions on rise time indicate that you still don't fully understand what slew rate limiting is and how it is defined either.
Also, I already told you one way possible way of reducing Cdom whilst maintaining stability - you increase the in value of the LPT emitter degeneration resistors.
If reducing Cdom or increasing the LTP standing current beyond a certain point does not increase the slew rate, then the VAS standing current WRT the (pre)driver input capacitance is likely the limiting factor.
hi Rail,
c33&34 look like they are in the wrong place.
q74 is upside down.
The input filters and NFB filters need to be reassessed.
Do something to separate the common CCS from both the LTP and Cascoded VAS.
L5 must be connected to pin3 of the soft start relay not pin5.
Change all the mains fuses to TxA (anti surge/time delay).
Change all the fuses after the smoothing to FxA (fast).
Delete all the fuses after the secondaries and before the smoothing.
c33&34 look like they are in the wrong place.
q74 is upside down.
The input filters and NFB filters need to be reassessed.
Do something to separate the common CCS from both the LTP and Cascoded VAS.
L5 must be connected to pin3 of the soft start relay not pin5.
Change all the mains fuses to TxA (anti surge/time delay).
Change all the fuses after the smoothing to FxA (fast).
Delete all the fuses after the secondaries and before the smoothing.
G.Kleinschmidt said:
Of course you are absolutelly wrong, because:
1) You are generalizing things 2) You do them complicated without significant reason.
The supply level can change dramaticaly the calculation of known parameters when increased in extreme values. Then the theories fall down and the only that remains is the trial & error proccess.
The VGS standing current it is about 20mA, and it is adjusted by observing a square wave on the scope so as the negative slewing has the same shape with the positive slewing. And 20mA for this stage are more than enough for 170Vpp signal level.
We have between us a big difference in our point of view. Generally you have the habit to make the things complex because you are examining those through schematics on the paper, calculations and i suppose also by using Spice simulation. Hahaha... i have tried this project with the EDspice simulator but the results proved far away from reality. If you are curious about the quality of my simulator, you can run to the site of Visionics for informations as well you must know that i run this application on a pure 64bit P.C. to avoid hung-ups. Instead my point of view it is most simplest, because i am examining a whole project by looking, for example, on the little secret filters arround it.
Provided that you can not find the parts that i mentioned above to minimize the rise time or to increase the extrinsic + intrinsic slew rate (hahaha, what a scientific terminology) let me to teach in you:
First of all, we should not see a tree and losing an entire forest. What i mean; you may consider the amplifier such as an operational amplifier. Then, the parts that remains outside, are those that forms the feedback circuit and the input filters. The secret it is finally the compensation filter R12-C9 around the feedback resistor R14. That is what regulates the rise time (or the slew limit as like the slew rate freaks). You can decrease the value of C9 and increase the value of R12 to obtain a smaller rise time of about 1,48ìsec, or to remove completelly this compensation network to obtain a rise time about 1ìsec. In the second case, because the overshoot appears again, you must increase the value of C13 (Cdom) to 120pF for eliminating it and the rise time will be finally about 1,2ìsec.
That is all!
Please let aside the lectures about slew rate, slew limit, extrinsic slew rate, intrinsic slew rate etc etc..
I can even live without them.Take it easy brother.
My projects operate just fine and heared very nice, without eximined slew rate. In the concrete project i have placed purposely the compensation network around the feedback resistor because it is an amplifier for P.A. use, so the abuse is given from the hand. In other projects for home use, usually i omit this compensation network.
Well, the free of charge lecture was finished for today. Tomorrow again, but on payment. (Hmmm...Because you need many lessons with an unbearable cost, i think that it is honest to make a small discount for you)
Regs
Fotios
fotios said:
Of course you are absolutelly wrong, because:
1) You are generalizing things 2) You do them complicated without significant reason.
The supply level can change dramaticaly the calculation of known parameters when increased in extreme values. Then the theories fall down and the only that remains is the trial & error proccess.
The VGS standing current it is about 20mA, and it is adjusted by observing a square wave on the scope so as the negative slewing has the same shape with the positive slewing. And 20mA for this stage are more than enough for 170Vpp signal level.
We have between us a big difference in our point of view. Generally you have the habit to make the things complex because you are examining those through schematics on the paper, calculations and i suppose also by using Spice simulation. Hahaha... i have tried this project with the EDspice simulator but the results proved far away from reality. If you are curious about the quality of my simulator, you can run to the site of Visionics for informations as well you must know that i run this application on a pure 64bit P.C. to avoid hung-ups. Instead my point of view it is most simplest, because i am examining a whole project by looking, for example, on the little secret filters arround it.
Provided that you can not find the parts that i mentioned above to minimize the rise time or to increase the extrinsic + intrinsic slew rate (hahaha, what a scientific terminology) let me to teach in you:
First of all, we should not see a tree and losing an entire forest. What i mean; you may consider the amplifier such as an operational amplifier. Then, the parts that remains outside, are those that forms the feedback circuit and the input filters. The secret it is finally the compensation filter R12-C9 around the feedback resistor R14. That is what regulates the rise time (or the slew limit as like the slew rate freaks). You can decrease the value of C9 and increase the value of R12 to obtain a smaller rise time of about 1,48ìsec, or to remove completelly this compensation network to obtain a rise time about 1ìsec. In the second case, because the overshoot appears again, you must increase the value of C13 (Cdom) to 120pF for eliminating it and the rise time will be finally about 1,2ìsec.
That is all!
Please let aside the lectures about slew rate, slew limit, extrinsic slew rate, intrinsic slew rate etc etc..
Take it easy brother.
My projects operate just fine and heared very nice, without eximined slew rate. In the concrete project i have placed purposely the compensation network around the feedback resistor because it is an amplifier for P.A. use, so the abuse is given from the hand. In other projects for home use, usually i omit this compensation network.
Well, the free of charge lecture was finished for today. Tomorrow again, but on payment. (Hmmm...Because you need many lessons with an unbearable cost, i think that it is honest to make a small discount for you)
Regs
Fotios
WTF?
You still don't understand (and obviously couldn't be bothered to understand) anything I said.
Your circuit topology is 99% D.Self. It has been analyzed to death by him and countless others. There are no mysteries as to how it operates. Your modified compensation is a kludge. And now you want to get personal?
This is becoming utterly inane and I certainly have nothing to prove to you.
Good bye!
I think that both of you ought to step back and avoid the personal. There's enough technical meat for an entire meal.
Since a suggestion didn't work, further action will have to be taken. I'm pulling some abusive posts and putting both of you guys in moderation.
An amplifier with infinite slew rate might still have a nonzero risetime if frequency response does not extend to infinity.
If the rising edge starts to bend towards the end like in fotios's graphs this slower rise is more probably due to frequency response than slew rate limiting. Increasing the slew rate limit in this case won't change rise time much.
If the rising edge starts to bend towards the end like in fotios's graphs this slower rise is more probably due to frequency response than slew rate limiting. Increasing the slew rate limit in this case won't change rise time much.
Hey!
Thanks for the answers!
megajocke
*
AndrewT
*
*
*
*
*
*
*
*
Again, thanks for replying!
Aq
Thanks for the answers!
megajocke
*
I'll test breaking capability before use, but what solution would you recommend?
AndrewT
*
You are absolutely correct, caps should be between ends of the bias circuit and the middle of feedback circuit. I will correct that. I made a mistake redrawing from mr. Slone's schematics.
*
Correct again, don't draw schematics when you are tired🙂
*
Can you suggest something for these filters? Input filtering is copied from Marantz PM8000 and feedback is fairly common. I really dont have a clue what could be better.
*
Again I must ask for some advice how to to that, I thought the LED reference could be fairly stiff to support both.
*
May you could specify the problem, because I don't see the error. When power is turned on, main transformer(L7) will get power through the resistors (L5) and when there is enough voltage on C3 relay will close and L7 will get it's power directly(L6).
*
Do you mean slow-blow fuses? I thought of using them everywhere.
*
This is seems better than using slow-blow, will do!
*
Do you mean fuses F1,F2,F3,F4,F7 and F8? Both Slone and Self give scematics in their books usng these fuses. Are they meaningless or bring something unwanted?
Again, thanks for replying!
Aq
one or both of the FxA fuses will blow on sustained overload if high DC voltage and high DC current maintain the arc across the relay contacts.raitraak said:
raitraak said:
The input filters (both High Pass and Low Pass) must set the bandwidth limits of the amplifier. The NFB filter should be at least half an octave lower than the input high pass filter
raitraak said:
Use, as a minimum, a resistor to separate the VAS CCS base from the LED reference, but better is to use two CCSs. However, completely separate CCSs don't like overload conditions. But, you can immobilise them with a thyristor on sustained overload as Crimson.
you're breaking the mains input circuit causing arcing across the contacts and then remaking the mains input with the second contact, just after the transformer has established it's flux but before the smoothing has charged up.raitraak said:
What you require is a shorting bypass (spst) across the resistor string, not a changeover relay (spdt).
Slowblow=anti-surge=time delay=TxA. Use them to start up transformers and motors.raitraak said:
Use fast fuses where the speed of interruption will allow normal operation of the circuit.raitraak said:
yes all 6 can be deleted. If you keep them they have to be rated to pass the charging transient that flows to the smoothing caps, making them much larger than the operational short term average current.raitraak said:
Good to see I am not the only one hitting it off with the
moderators. 🙂
I always make comparisons about how Quasi=D.self
DX= 20 year old topology.... get arguments!!
Most of the amps on DIYaudio are based on old designs and
Nuubs think they are groundbreaking super duper
developments.(and argue to defend thread creators like gods.)
I am thankful for the resources available at DIY but
I don't think any are true intellectual property that
can't be openly criticized in the wild.
Can't wait to create my own thread to have it "torn apart"
by the "vultures". I would take it as constructive criticism
in which to improve my design.
The"power amp under development
thread" (D. Self design) is a prime example of this.😀
OS
moderators. 🙂
I always make comparisons about how Quasi=D.self
DX= 20 year old topology.... get arguments!!
Most of the amps on DIYaudio are based on old designs and
Nuubs think they are groundbreaking super duper
developments.(and argue to defend thread creators like gods.)
I am thankful for the resources available at DIY but
I don't think any are true intellectual property that
can't be openly criticized in the wild.
Can't wait to create my own thread to have it "torn apart"
by the "vultures". I would take it as constructive criticism
in which to improve my design.
The"power amp under development
thread" (D. Self design) is a prime example of this.😀
OS
I want to take the opportunity to maybe discuss the q59 / q36 (emitter to collector connection).
In my experiments with this kind of connection I always had to put a small resistor(1-10ohm) between the transistors, emitter to collector in order for it not to be prone to oscillations, even when just connecting them by them selves ,just experimenting..
Now why is that, are there any special characteristics for Q59 /Q69 that is needed, does it need to be especially slow (or fast).
or is it hfe related, or am I missing something simple..
(this will not show in multisim, and possible not in the final circuit, but sometimes I just like taking a a couple of components and experiment a little just by themselves just for fun, maybe at lunch at work or someting)..
Also , should'nt this circuit really be possible to make stable enough not to have to have such a output coil? or is it already.. I dont want to raise that old discussion again but I do wonder If it's possible to ask the question: is it better to make a less stable circuit and use a bigger output coil, or build stability into the circuit and use no coil.
,,,
also about the risk of copying someone else, It's hard to come up with something really new but as I see it, for example most cars are very alike, still some of us prefer a special type (toyota celica supra rocks ;-D ). I did do something (possably somewhat stupid) about that, I collected as many power amp schematics that I could and sorted them in categories, then began experimenting in the area where there were no schematics 😛 or at least not many., that way, even if I fail, it's going to be more interresting hehehe
( but then again, if I happen to come across a nice mitsubishi starion that would be nice to have too)..
ps. just because I like old cars, does'nt mean I use germanium transistors, but I do like tubes sometimes, if they're mated with mosfets..
In my experiments with this kind of connection I always had to put a small resistor(1-10ohm) between the transistors, emitter to collector in order for it not to be prone to oscillations, even when just connecting them by them selves ,just experimenting..
Now why is that, are there any special characteristics for Q59 /Q69 that is needed, does it need to be especially slow (or fast).
or is it hfe related, or am I missing something simple..
(this will not show in multisim, and possible not in the final circuit, but sometimes I just like taking a a couple of components and experiment a little just by themselves just for fun, maybe at lunch at work or someting)..
Also , should'nt this circuit really be possible to make stable enough not to have to have such a output coil? or is it already.. I dont want to raise that old discussion again but I do wonder If it's possible to ask the question: is it better to make a less stable circuit and use a bigger output coil, or build stability into the circuit and use no coil.
,,,
also about the risk of copying someone else, It's hard to come up with something really new but as I see it, for example most cars are very alike, still some of us prefer a special type (toyota celica supra rocks ;-D ). I did do something (possably somewhat stupid) about that, I collected as many power amp schematics that I could and sorted them in categories, then began experimenting in the area where there were no schematics 😛 or at least not many., that way, even if I fail, it's going to be more interresting hehehe
( but then again, if I happen to come across a nice mitsubishi starion that would be nice to have too)..
ps. just because I like old cars, does'nt mean I use germanium transistors, but I do like tubes sometimes, if they're mated with mosfets..
raitraak said:
You have done a tremendous work on this 300 Watt project.
I have downloaded your all circuits.
The power amplifier is the only i have studied, a bit.
To me, it looks good.
Choice of transistors could NOT be better, if we are talking Diy Project 😉
Even for a professional power amplifier those devices are very good.
---------------
Please, raitraak 😎
Keep us informed about the develoopment
into one working 300 Watt RMS beast!
Audio Regards
Lineup
Hey!
Thank You, AndrewT for being so patient!🙂
AndrewT
NFB: I actually don't know how to calculate this filter and cannot find any examples. If using the usual equation f = 1 / (2*Pi*R*C), then the value seems incorrect (100MHz for 500ohm and 5pF). Can you give a link or an example how this is done.
ostripper
The reason I like books from Self and Slone, is that they explain the main topologies very clearly. Of course they cannot explain everything, it takes a LOT of pages to do that. When you have very little time, and you want to build something, that you can use in real life and in first try, it's a way to go.
nikwal
Lets see what others think of the need of this resistor, but when stability is at stake, adding a resistor or two is not a problem.
Lineup
I'll post every development here🙂
Updated schematics:
MAIN
PWR
SOFT_START
Some questions from previous post:
Regards,
Aq
Thank You, AndrewT for being so patient!🙂
AndrewT
So I'll leave things, like they are now.
Input filter: Low-Pass filter 50kHz should be ok? With High-Pass there is a problem, I don't want the input resistance to be lower than 10k, so I would need a bigger cap. This means using electrolytic one. I already have the 20uF non-polar, so I'd like to stick with this configuration(0,8Hz), is it ok?
NFB: I actually don't know how to calculate this filter and cannot find any examples. If using the usual equation f = 1 / (2*Pi*R*C), then the value seems incorrect (100MHz for 500ohm and 5pF). Can you give a link or an example how this is done.
Added resistors for both transistors. Is 1k ok?
As always, this seems more reasonable.🙂
Deleted them.
ostripper
The reason I like books from Self and Slone, is that they explain the main topologies very clearly. Of course they cannot explain everything, it takes a LOT of pages to do that. When you have very little time, and you want to build something, that you can use in real life and in first try, it's a way to go.
nikwal
Lets see what others think of the need of this resistor, but when stability is at stake, adding a resistor or two is not a problem.
Lineup
I'll post every development here🙂
Updated schematics:
MAIN
PWR
SOFT_START
Some questions from previous post:
Regards,
Aq
- Status
- Not open for further replies.