A newbie question, In this amplifier I do not understand how Q4, Q5, ZD2, R3, R4, R5 works, I believe it's a current sink but I'm not sure how it works, would like some matematical explanations.
The voltage gain stage used is a differential pair right?
Thanks for any help
Diogo
The voltage gain stage used is a differential pair right?
Thanks for any help
Diogo
Q4 and Q5 forms a cascode. You can read more about it at... can't reach www.passdiy.com at the moment.
The purpose is:
* Make the input stage faster (5-10 times!)
* Withstand higher voltage
Q4 and Q5 are base grounded stages. The collector load of Q2 is more or less zero (diod plus a zener down to ground = low impedance) therefore the "Miller" effect is small. This effect slows down the transistor.
The cascode has no effect of gain nor input noise. This two transistor circuit forms a "super transistor", fast, low noise, high voltage, high Hfe (usually)
The purpose is:
* Make the input stage faster (5-10 times!)
* Withstand higher voltage
Q4 and Q5 are base grounded stages. The collector load of Q2 is more or less zero (diod plus a zener down to ground = low impedance) therefore the "Miller" effect is small. This effect slows down the transistor.
The cascode has no effect of gain nor input noise. This two transistor circuit forms a "super transistor", fast, low noise, high voltage, high Hfe (usually)
In other words, Q4 and Q5 just regulate the voltage at Q2 and Q3 collectors to stop their Vce rating being exceeded - as the psu voltage is 70V (BC549 are rated at 65V). The current from Q2 and Q3 just passes straight through Q4 and Q5.
I don't think the speed of the input stage is much affected by Q4 and Q5 because the collector loads of 2.2k are small, so the miller effect is not so important. In fact, usually you don't want the input stage too be too fast because it affects the stability of the amp. However, you do get capacitive coupling across the be of the transistors which has not been decoupled to ground in this design. For better performance I'd consider removing Q4 and Q5 and replacing them with two zeners.
I don't think the speed of the input stage is much affected by Q4 and Q5 because the collector loads of 2.2k are small, so the miller effect is not so important. In fact, usually you don't want the input stage too be too fast because it affects the stability of the amp. However, you do get capacitive coupling across the be of the transistors which has not been decoupled to ground in this design. For better performance I'd consider removing Q4 and Q5 and replacing them with two zeners.
Ok I think I undertand Why and how, but I still whant to know how to calculate this cascode stage voltage and currents, the resistor values and the zener voltage.
Thank you guys!
Thank you guys!
traderbam said:In fact, usually you don't want the input stage too be too fast because it affects the stability of the amp.
My point of view is quite the opposite! What a fast input stage you will only get one dominant pole = easier to feedback => better stability.
This is especially true if you want to make a high slew rate amp. The other poles are far far away and can't make no harm.
"This is especially true if you want to make a high slew rate amp"
Why do you want a high slew rate amp? The highest frequency being amplified is 20kHz isn't it? What slew rate do you consider acceptable?
Why do you want a high slew rate amp? The highest frequency being amplified is 20kHz isn't it? What slew rate do you consider acceptable?
Slow amps tends to have high distortion in the higher frequencies.
I think it's wise make an amp with a power bandwidth of > 100 kHz but > 300-400 kHz aren't neccessary.
I think it's wise make an amp with a power bandwidth of > 100 kHz but > 300-400 kHz aren't neccessary.
traderbam said:What slew rate do you consider acceptable?
The slew rate is determined by max output voltage and max frequency. 1000 W amp needs much more slew rate than a 15 W amp.
But if we talk numbers, 50 V/µs for a 200-300 W amp is normal.
I believe that the formula for slew rate for a sinewave is:
SL= 2.pi.f.Vpeak in volts/sec
Where f is the sinewave frequency in Hz and Vpeak is the peak amplitude of the sinewave.
Eg: for a 40V, 20kHz sinewave a SL of about 5V/us is required. I'd say this is a typical requirement for home amps.
The bandwidth needs to be at least 20kHz. Why does it need to be any higher? After all, the bandwidth of your CD player is somewhere between 20 and 22kHz. Why should an amp sound better when it can amplify out beyond 100kHz? (except perhaps to dogs and bats). Wouldn't it be better if it rolled off at 20kHz so as not to amplify anything it doesn't need to - wouldn't this put less strain on the amp and less distortion?
SL= 2.pi.f.Vpeak in volts/sec
Where f is the sinewave frequency in Hz and Vpeak is the peak amplitude of the sinewave.
Eg: for a 40V, 20kHz sinewave a SL of about 5V/us is required. I'd say this is a typical requirement for home amps.
The bandwidth needs to be at least 20kHz. Why does it need to be any higher? After all, the bandwidth of your CD player is somewhere between 20 and 22kHz. Why should an amp sound better when it can amplify out beyond 100kHz? (except perhaps to dogs and bats). Wouldn't it be better if it rolled off at 20kHz so as not to amplify anything it doesn't need to - wouldn't this put less strain on the amp and less distortion?
You forget that you get nasty distortion due to slew rate limiting. If you have low feedback you won't get so much slew rate limiting but you will instead get higher harmonic distortion.
Considering what you say, why do we need 96 kSp or even 192 kSp?
OK, I talking about my taste. I don't settle for an amp that barely can make 20-20000 Hz.
You are totally right in questioning about the minimum performance.
Considering what you say, why do we need 96 kSp or even 192 kSp?
OK, I talking about my taste. I don't settle for an amp that barely can make 20-20000 Hz.
You are totally right in questioning about the minimum performance.
What about my question guys?
And I agree with peranders, a fast amp sounds mutch better,
try to amplify eletronic music like trance ,with a faster amp and listen to it, I don't have to use the audio analyser to be sure of that!
And I agree with peranders, a fast amp sounds mutch better,
try to amplify eletronic music like trance ,with a faster amp and listen to it, I don't have to use the audio analyser to be sure of that!
cascoded diff pairs
I think we can add one more reason to use cascodes, namely
that it will keep the Vce of the diff pair transistors almost
constant, thus minimizing the Early effect. Or is this considered
a minor reason, not worth the trouble?
I think we can add one more reason to use cascodes, namely
that it will keep the Vce of the diff pair transistors almost
constant, thus minimizing the Early effect. Or is this considered
a minor reason, not worth the trouble?
Re: cascoded diff pairs
This is why distortion gets reduced and the speed increased.... this isn't minor reason, this is the basics of the whole stage.
Christer said:I think we can add one more reason to use cascodes, namely
that it will keep the Vce of the diff pair transistors almost
constant, thus minimizing the Early effect. Or is this considered
a minor reason, not worth the trouble?
This is why distortion gets reduced and the speed increased.... this isn't minor reason, this is the basics of the whole stage.
I missed that you mentioned lower distorsion in a follow-up post,
but you didn't mention the cause for it. It seems we agree that
the cause is increased linearity by minimizing the Early effect.
I just wanted to check that I have the correct understanding of
cascoding, and it seems I have (at least, until somebody else
comes around having a different opinion from yours 🙂 ).
but you didn't mention the cause for it. It seems we agree that
the cause is increased linearity by minimizing the Early effect.
I just wanted to check that I have the correct understanding of
cascoding, and it seems I have (at least, until somebody else
comes around having a different opinion from yours 🙂 ).
Cascodes
We are getting into uneasy ground here, as some respected designers like Nelson Pass or Hugh Dean do not use them, favouring a simpler resistor replacement there.
As I couldn't yet carry on my own tests there, replacing cascodes and constant current sources with resistors and see how they sound, that looks like an interrogation area for me.
My present amplifier uses both cascodes and ccs, so I think I'm familiar with their sound. Now I want to listen to the simpler designs and see if there's an issue there.
Can't say anything on current mirrors compared to cascodes though, whether one or th eother should be preferred.
Other opinions, based on actual listenings, should be welcome.
Carlos
We are getting into uneasy ground here, as some respected designers like Nelson Pass or Hugh Dean do not use them, favouring a simpler resistor replacement there.
As I couldn't yet carry on my own tests there, replacing cascodes and constant current sources with resistors and see how they sound, that looks like an interrogation area for me.
My present amplifier uses both cascodes and ccs, so I think I'm familiar with their sound. Now I want to listen to the simpler designs and see if there's an issue there.
Can't say anything on current mirrors compared to cascodes though, whether one or th eother should be preferred.
Other opinions, based on actual listenings, should be welcome.
Carlos
Re: Cascodes
If I don't remember wrong Nelson Pass uses cascodes if they are put in the right places. Hadn't he some patent also?
Current mirrors and cascodes are two different things. Cascodes haven't any effect on gain which current mirrors have if they are collector loads.
carlmart said:We are getting into uneasy ground here, as some respected designers like Nelson Pass or Hugh Dean do not use them, favouring a simpler resistor replacement there.
If I don't remember wrong Nelson Pass uses cascodes if they are put in the right places. Hadn't he some patent also?
Current mirrors and cascodes are two different things. Cascodes haven't any effect on gain which current mirrors have if they are collector loads.
The sound of things
If I don't remember wrong Nelson Pass uses cascodes if they are put in the right places. Hadn't he some patent also?
Current mirrors and cascodes are two different things. Cascodes haven't any effect on gain which current mirrors have if they are collector loads. [/B][/QUOTE]
First of all, what I'm referring to is how certain arrangements may influence the sound quality. Whether they are current mirrors, cascodes or ellaborated constant current sources.
Second, if I'm not wrong Nelson Pass does not use cascodes on any of the Alephs that many DIYers are building, and Pass did comment on the advantages of simpler circuits.
The only Pass design I've seen that does use cascodes is the A75, though there are certainly others I haven't seen.
Carlos
If I don't remember wrong Nelson Pass uses cascodes if they are put in the right places. Hadn't he some patent also?
Current mirrors and cascodes are two different things. Cascodes haven't any effect on gain which current mirrors have if they are collector loads. [/B][/QUOTE]
First of all, what I'm referring to is how certain arrangements may influence the sound quality. Whether they are current mirrors, cascodes or ellaborated constant current sources.
Second, if I'm not wrong Nelson Pass does not use cascodes on any of the Alephs that many DIYers are building, and Pass did comment on the advantages of simpler circuits.
The only Pass design I've seen that does use cascodes is the A75, though there are certainly others I haven't seen.
Carlos
- Status
- Not open for further replies.
- Home
- Amplifiers
- Solid State
- Symmetrical 400W amplifier