| dododio |
Hello,
Everything is in the subject: i try to build a low frequencies preamplifier with a differential structure (40dB, input voltage swing: from 0.01V to 0.1V and a good THD over the entire bandwith(audio)).
I simulate my circuits in ORCAD Pspice.
With a simple diff pair with current sink followed by a class A amplifier DC coulped with the previous stage (for the feedback) i can't lower the thd specialy when the input voltage rise and i would like to get a good thd over the dynamic of the signal.
When Vce swing Ic swing too so béta swing too=>distortion,
I heard somewhere that cascode could minimize the third order distortion is that right?, is there any other solution?
I also heard about cross coupled amplifier...
If somebody can bring some theory about linearisation of the bjt diff amp...
Sincerely,
dd |
|
|
| nigelwright7557 |
| quote: | Originally posted by dododio
Hello,
When Vce swing Ic swing too so béta swing too=>distortion,
dd |
Hwo about using a constant current source to keep the current the same ? |
|
|
| YFW |
| You should put negative feedback into the circuit to reduce the THD. |
|
|
| PMA |
You have to make a choice of better input topology.
Cheers,
PM |
|
|
| destroyer X |
and adjust your impedance using dummy resistances into the input as loads to match your pré amplifier...keep the signal inside the limits and distortion will be something you will not even listen.
Of course measurements instruments will see distortions and will read them..... but you will not perceive...so... not your problem...the problem belongs to the Measurement instrument...it may be disturbed or bothered...you will remain very happy listening good sound.
Big secret... and for free... adjust your non distorted level to 75 percent of your volume.... you will have room to increase low level from your audio source and will not have too much risk operating confortable audition levels.
Distortions will always exists....and do not bother about the amplifiers...the speaker will be distorting 100 percent more than the amplifiers...so.... the amplifier is something that is not the real problem to create worries.
Watch the speaker, that dam thing...the air pump... awfull mechanics... very ancient thing.... the guilty is there!
Give a shot into your 6 volts RMS output pré amplifier...not needed, problem creator... the audio sources already has level enougth to drive any power amplifier to the maximum level of power..so....pré amplifiers is one more thing to create problems, as it is inside the audio chain...use potenciometers to control the input volume and be happy my friend.
regards,
Carlos |
|
|
| nigelwright7557 |
| quote: | Originally posted by destroyer X
Watch the speaker, that dam thing...the air pump... awfull mechanics... very ancient thing.... the guilty is there!
regards,
Carlos |
People spend ages getting the amp right then send the signal to a speaker with a wildy fluctuating frequency curve. |
|
|
| ttan98 |
| quote: | Originally posted by nigelwright7557
People spend ages getting the amp right then send the signal to a speaker with a wildy fluctuating frequency curve. |
Looks like to don't visit the loudspeaker forum often enough... |
|
|
| nigelwright7557 |
| quote: | Originally posted by ttan98
Looks like to don't visit the loudspeaker forum often enough... |
Not at all, I just read the loudspeaker specification pdf's !!! |
|
|
| lineup |
| quote: | Originally posted by dododio
Hello,
- low frequencies preamplifier
- with a differential structure
- (40dB,
- input voltage swing: from 0.01V to 0.1V and
- a good THD over the entire bandwith(audio)).
I simulate my circuits in ORCAD Pspice.
- a simple diff pair with
- current sink
- followed by a class A amplifier DC coulped with the previous stage (for the feedback)
i can't lower the thd specialy when the input voltage rise and i would like to get a good thd over the dynamic of the signal.
When Vce swing Ic swing too so béta swing too=>distortion,
I heard somewhere that cascode could minimize the third order distortion is that right?, is there any other solution?
I also heard about cross coupled amplifier...
If somebody can bring some theory about linearisation of the bjt diff amp...
Sincerely,
dd |
When I know too little, I do not want to join in doing speculations.
I have several questions, for clarification, if you can not post a Drawing / Schematic,
to let us see what is going on.
------------------------
1. Power supply? Voltage?
I suppose dual, negative + postive with 0 Volt in middle.
2. 40 dB .. what?
If you by this mean x100 v-gain, maybe
3. LOAD?
What kind of load you intend to drive with preamp?
1 kohm .. or at worst case 5 or 10 kohm
4. What output max voltage you need across this load?
Using 0.01 Vrms / 0.1 Vrms .. at gain x100 = 1 - 10 Vrms Output!!!
This means a peak voltage out = 1.414 x 10 = +- 14.14 Volt.
Would require, in my book, at least a +- 20 Volt supply.
Preferably, a well regulated supply.
This is what good preamplifier deserves.
5. Not many devices fed from a preamp need more than 1- 2 Vrms input.
As hinted by destroyer X, already.
----------------------
I look forward to get mor info.
Then I would give you some good advices, how to do.
And how to lower your THD a bit. For good audio sound ( maybe even Hi-Fi :) ).
As the others posters here have good knowledge to advice, as well.
Regards, Lineup |
|
|
| nigelwright7557 |
| quote: | Originally posted by lineup
When I know too little, I do not want to join in doing speculations.
Regards, Lineup |
And how I am supposed to learn if I dont join in and ask questions !!!!!
I have been in electronics for 32 years so I hardly a novice !
You need to take a long hard look at a speaker datasheet before having a dig at me.
The Fane Collossus has more curves than Pamela Anderson !!!
|
|
|
| nelsonvandal |
| Why don't you show us your schematic? |
|
|
| nigelwright7557 |
| quote: | Originally posted by nelsonvandal
Why don't you show us your schematic? |
Enclosed........... |
|
|
| homemodder |
Hi Guys
I think what dododio is looking for is somewhat more technical. The crosscoupled design he mentions does work to a certain extent but has its drawbacks there is a thread about it here on the forum and you could search for it. Cascoding does help to a degree as does the use of cfp.
Dododio if you really want to get into it i can help you on the way, but im a terrible teacher and i dont have much time so youll have to read up on it by yourself. So I suggest you to look up the patents by Patric Quin 4,146,844 and the later improved versions by James Woo 4,779,057. Strangely enough there is only one person that has incorperated these techniques in a amp and i can bet that amp sounds damm fine. Perfect for preamps. I think there be some stability issues but theres a solution to every problem. If you have problems getting this info drop me a email and ill send it to you. I havent got the time to try or look into it with care, my wife will kill me and my hobbies :bawling: :D.
Alex |
|
|
| dododio |
Hello,
here is the schematic,
the signal level are amplitude not rms,
input generator impedance :22kohms,
output impedance :600ohms after a buffer stage not drawn.
sincerely,
dd |
|
|
| homemodder |
Hi Forr
Who is textronix?? Ones using Woo inventions??
I have read the sites you mentioned. It is good yes and i thought of mentioning it as it is simple and effective. Ive used and seen cfp cascoded designs even before i was aware of perrot. We did experiments at varsity concerning linearity and this type of design our professor showed us. Never looked at it from a memory effect perspective but from a linearity one, i was in varsity a long time ago, before perrots patent. The two go hand in hand as i see it.
A shootout would be interesting between the two. There is also now a third very clever and simple design by i think his name is kullish, using error correction that can be brought into the mix. Someone that comes to this forum has designed a circuit based on woo s ideas that looks like it will outperform woo s original and this person i am pretty sure knows the perrot papers. I am not so sure perrots circuit would come out on top, this offcourse is from a THD perspective as there are other matters too especially in audio. The only one i have listened to is perrots circuit.
Alex |
|
|
| unclejed613 |
| you don't seem to have any global feedback, which is really important with a diff amp. another possible problem is your "current source" which looks like it's operating more like an "amplified" resistor. a better approach would be to dump the "mirror" transistor in the current source and replace it with a pair of diodes. the way a current source works is that it sets up a reference voltage at the base of the CS transistor, usually with a pair of diodes or an LED. the transistor tries to maintain one diode drop across the B-E junction, while dropping the other diode drop across the emitter resistor. the voltage across the emitter resistor reduces the B-E current until it stabilizes at a preset current, and the current source is in equalibrium. any attempts to change that current (such as changing the voltage on the collector) will cause the transistor to maintain equalibrium, which keeps the current constant. |
|
|
| dododio |
Hi,
forr, thank for your links,
unclejed 613, i forget to connect the B&C for the current mirror and there is a negative global feedback formed by resistor after class a common emiter stages.
Another question is : if the transistor is current feed then the distortion comming from Ib(Vbe)(with the exponnential relationship) will be cancelled? => if this is right this kind of attack could support a greater voltage input swing than a classical voltage attack (so with less distortion)?
So there is 2 kind of distortion (dependant on amplitude)?
sincerely,
ch |
|
|
| DRC |
| Your EF output stage is not inside the feedback loop. If the output swing is significant, compared to the supply rails, then this will create some distortion. It could be improved with a current source - if you want to keep it outside the feedback loop. |
|
|
| forr |
| quote: | Originally posted by homemodder
Hi Forr
Who is textronix?? Ones using Woo inventions??
I have read the sites you mentioned. It is good yes and i thought of mentioning it as it is simple and effective. Ive used and seen cfp cascoded designs even before i was aware of perrot. We did experiments at varsity concerning linearity and this type of design our professor showed us. Never looked at it from a memory effect perspective but from a linearity one, i was in varsity a long time ago, before perrots patent. The two go hand in hand as i see it.
A shootout would be interesting between the two. There is also now a third very clever and simple design by i think his name is kullish, using error correction that can be brought into the mix. Someone that comes to this forum has designed a circuit based on woo s ideas that looks like it will outperform woo s original and this person i am pretty sure knows the perrot papers. I am not so sure perrots circuit would come out on top, this offcourse is from a THD perspective as there are other matters too especially in audio. The only one i have listened to is perrots circuit.
Alex |
Hi Homemolder
Tektronix is a famous maker of osccilloscopes.
Perrot measured many other kinds of differential input stages, many were better than the Cascomp. I can provide you details, just email me (in french but figures are easy to understand).
Perrot prefered circuit should not be seen as a variation of the Sziklai pair neither as using a cascode stage. Its input bipolar is loaded by a CCS and is followed by a complementary PNP - NPN with coupled emitters which is a series differential stage, also called a Rush circuit. A common base transistor does not mean that a circuit is necessarily a cascode. |
|
|
| homemodder |
Hi Forr
I would be very interested in those articles, thank you. Half my family live in France and will be here on holidays soon and can translate for me. Its always easier for me to understand topologies looking at simplified circuit diagram as their are sometimes discripencies as how we view a topology. Interesting you mentioning rush cascodes, have played a bit with these but didnt like the sound, the same goes for kulish that i tried earlier today. Will try some more experiments as sometimes its in the implementation. Excellent results can be obtained with the kulish, its a shame about the sound.
In my view perrots circuit does not fully address the Vbe temperature dependance. I havent tried cascomp circuit as it would require a lot of time as i think there are gonna be some difficult stability issues with it. However this new circuit ive seen does not seem to have any problems at all, the very little people that have listened to it say its brilliant, people that are familiar with high end equipment.
My email is alexmiguel33@hotmail.com
Thanks |
|
|
| Lumba Ogir |
homemodder,
well, linearization is not quite a cleanup sweeping out the trash, it`s more about shaping the sound to the ears liking with remaining distortions. When lowering certain distortions, inevitably other ones will be created so many of those circuits do more damage than good, their sonic impact has to be investigated. I know you do that thoroughly, a rare phenomenon these days, people rather prefer to fool around with simulations to find out sound quality...a terrible mistake... |
|
|
| lineup |
| quote: | Originally posted by dododio
here is the schematic,
the signal level are amplitude not rms,
input generator impedance :22kohms,
output impedance :600ohms after a buffer stage not drawn. |
| quote: | Originally posted by DRC
Your EF output stage is not inside the feedback loop.
If the output swing is significant, compared to the supply rails,
then this will create some distortion.
It could be improved with a current source
- if you want to keep it outside the feedback loop. |
Referng to your schematic, dododio
http://www.diyaudio.com/forums/atta...tamp=1215109780
The distortion level at the output transistors BASE
is dependent of your degeneration resistors,
BOTH in the input pair and in those PNP.
Lower gain per transistor/stage (higher resistor values) = lower distortion.
More transistors/stages gives more gain and more feedback factor = lower distortion.
As DRC pointed out,
additional distortion is added by the output transistors.
This addition is usually very low, even compared to one Feedback corrected Voltage amplifier.
And will be even lower if using CCS loaded output
and high input impedance for your output devices.
THD gets higher the more load(lower impedance) and the higher signal amplitude.
You adjust the load of your stages + the output to a suitable, optimized load
to get the output you want/need.
And hopefully by this you have as low distortion you can get
from your used circuit.
It is quite a nice amplfier you use, actually :)
Lineup |
|
|
| unclejed613 |
| the two PNP transistors are acting as transconductance amps, which means they're taking a current input (from the diff pair) and turning it into a voltage. the voltage waveform on the bases of the PNPs will be a very distorted voltage waveform, but since they're being driven by current, that doesn't matter. the collector resistors for the PNPs should be replaced by current sources, which would lower distortion a lot. you would also improve performance a lot by using a pair of complementary teansistors on the output stage, biased into class A operation. once that is done, this will resemble a bridged amp. |
|
|
| lineup |
current sources
will improve most any amplifier .. at least in terms of THD distortion
sometimes they are NECESSARY .. to get a fair amplifier stage
other times we can use bootstrapping or plain resistors instead
with evenso good result
think unclejed is right :)
this one case is where current sources would make a whole lot of difference
:cool: for the better |
|
|
|
