Q14 and Q15
Even without biasing of the bases of Q14 and Q15, these transistors still act as regulators. They have only limited collector-emitter voltage, though. For no ripple on the supply line the collector will be 0.7V more positive than the emitter (for Q14). That means that the supply may go down approx 0.7V before Q14 will go into saturation, and 0.7V is not much for ripple on the supply of a power amp. But as long as the ripple is less than that, Q14 will regulate and lower the noise/ripple to the pre-stage. If the supply goes down more, the collector base diode of Q14 may go into conduction and C6 will be discharged, and consequently C2 will not be charged anymore and will loose voltage after a while, dependant on the value of C2.
I think I would put protection diodes from the emitter to the collector of Q14 (and Q15); anode to E(Q14) and cathode to C(Q14), for Q15 the other way around. These will protect Q14 from being damaged if the supply goes down quickly and C2 is still charged. The emitter base diode of Q14 will zener, the base collector diode will conduct and C2 is discharged to the supply via Q14. Q14 may not survive, depending on transistor type and value of C2. For Q15 a similar story.
Steven
Even without biasing of the bases of Q14 and Q15, these transistors still act as regulators. They have only limited collector-emitter voltage, though. For no ripple on the supply line the collector will be 0.7V more positive than the emitter (for Q14). That means that the supply may go down approx 0.7V before Q14 will go into saturation, and 0.7V is not much for ripple on the supply of a power amp. But as long as the ripple is less than that, Q14 will regulate and lower the noise/ripple to the pre-stage. If the supply goes down more, the collector base diode of Q14 may go into conduction and C6 will be discharged, and consequently C2 will not be charged anymore and will loose voltage after a while, dependant on the value of C2.
I think I would put protection diodes from the emitter to the collector of Q14 (and Q15); anode to E(Q14) and cathode to C(Q14), for Q15 the other way around. These will protect Q14 from being damaged if the supply goes down quickly and C2 is still charged. The emitter base diode of Q14 will zener, the base collector diode will conduct and C2 is discharged to the supply via Q14. Q14 may not survive, depending on transistor type and value of C2. For Q15 a similar story.
Steven
Q14 and Q15 is capacitance multiplier. The basic equation is just simple, how many current you wanted to pass Q14 and Q15, divide it by Hfe, and there you have the C-B resistor. You can put zener to ground, but it will be zener regulator, not cap multiplier.
It acts to multiply the capacitance, so there is capacitance from base to ground.
It acts to multiply the capacitance, so there is capacitance from base to ground.
http://diyaudio.com/forums/showthread.php?postid=394464#post394464
Hi...Lumanaw
The different between pre and power amplifier basic design are 1. bias current at driver stage to achieve the stability at desired bandwith and
2. most linear region transistor operation at high signal level
3. capability to drive lower impedance load..
Cheers,,,
Lukio
Hi...Lumanaw
The different between pre and power amplifier basic design are 1. bias current at driver stage to achieve the stability at desired bandwith and
2. most linear region transistor operation at high signal level
3. capability to drive lower impedance load..
Cheers,,,
Lukio
PRR said:
Hi prr...you've become my all-time favovourite commentator on this forum with nuggets such as this.....cheers!
Hi Mr Curl...
What advantages have you located with respect to the fully complementary (visually!) voltage gain stage vis a vis the generic topology in audio band applications??
john curl said:
What advantages have you located with respect to the fully complementary (visually!) voltage gain stage vis a vis the generic topology in audio band applications??
hi pavel
This is not an advantage, as you can have symmetrical slew rates with the conventional design...subject to modifications....eg bootstrap....
PMA said:
This is not an advantage, as you can have symmetrical slew rates with the conventional design...subject to modifications....eg bootstrap....
I have been working on a DIY Headphone Amp using a similar Folded Cascode topology and I find mine has a lot in common with this
A lot of researching reveals also that this is a popular implementation of Folded Cascode Amplifiers I see similarities in these circuits to early Solid state Luxman Amps, however As expected each designers implementation of this idea while generically similar are unique in individual details. One classic variation is if a Diff amp is included in some designs such as the design I just cited, while others dispense with this additional diif amp in the second stage as is done in this Implementation,
in addition other subtitle variations include to current source actively or passively along with other housekeeping duties. Feedback and compensation are also critical. This brings me to my question Referring Back to the first design the Load Resistors for the vas stage design the 47k to ground from the Collector and emitter of the VBE multiplier. This appears to be the most popular method however I was thinking of returning them to the inverting input sum node along with the Overall loop feedback loop. ( Again I cite this design)
I would like your views on these two methods as I hope to be able to make my design stable without any compensation capacitors and yet retain a comfortable Amount of overall feedback. The second method appears to have the advantage of using the vas second stage load resistors as an additional local feedback loop around just the Voltage gain section and thus not have to deal with the excessive phase shift of the output stage. I understand a simple solution is a cap from the collector of the Vbe multiplier to the inverting input nd in effect rolling off the bandwidth of the vas stage to a point the output stage can deal with. In my version I am using a three stage Darling connected complementary output so stability is an issue with my wish list however I want to add to that list no frequency compensation capacitors be required.
A lot of researching reveals also that this is a popular implementation of Folded Cascode Amplifiers I see similarities in these circuits to early Solid state Luxman Amps, however As expected each designers implementation of this idea while generically similar are unique in individual details. One classic variation is if a Diff amp is included in some designs such as the design I just cited, while others dispense with this additional diif amp in the second stage as is done in this Implementation,
in addition other subtitle variations include to current source actively or passively along with other housekeeping duties. Feedback and compensation are also critical. This brings me to my question Referring Back to the first design the Load Resistors for the vas stage design the 47k to ground from the Collector and emitter of the VBE multiplier. This appears to be the most popular method however I was thinking of returning them to the inverting input sum node along with the Overall loop feedback loop. ( Again I cite this design)
I would like your views on these two methods as I hope to be able to make my design stable without any compensation capacitors and yet retain a comfortable Amount of overall feedback. The second method appears to have the advantage of using the vas second stage load resistors as an additional local feedback loop around just the Voltage gain section and thus not have to deal with the excessive phase shift of the output stage. I understand a simple solution is a cap from the collector of the Vbe multiplier to the inverting input nd in effect rolling off the bandwidth of the vas stage to a point the output stage can deal with. In my version I am using a three stage Darling connected complementary output so stability is an issue with my wish list however I want to add to that list no frequency compensation capacitors be required.
Hi, PPI,
I've got (bad) experience with folded cascode. If your output stage is classA or having error correction scheme, you can use folded cascode and have good result.
I made low biased classAB without error correction, and the front end is folded cascode, it's just not working. OL gain is lack too much, the sound is like all treble to the music.
It is right that folded cascode (ie : low OL gain) can dispense stability caps like miller cap, and with folded cascode usually stabilizing is done by 5pf-20pf cap from collector of folded cascode to the inverting input, or even no cap at all, but the main problem is how many OL gain minimum is required to fix all the bad things in the whole amp? Sometimes folded cascode just don't provide enough OL gain to be burned for good CL gain performance.
Other method to linearize OL curve (beside putting R from collector of VAS to ground) is to put 220k in the place of miller cap (base-collector of VAS) and this works good too 😀 But my experience, load to ground or 220k in VAS local loop does not affect stability much, the best way to stabilize an amp is still put caps.
I've got (bad) experience with folded cascode. If your output stage is classA or having error correction scheme, you can use folded cascode and have good result.
I made low biased classAB without error correction, and the front end is folded cascode, it's just not working. OL gain is lack too much, the sound is like all treble to the music.
It is right that folded cascode (ie : low OL gain) can dispense stability caps like miller cap, and with folded cascode usually stabilizing is done by 5pf-20pf cap from collector of folded cascode to the inverting input, or even no cap at all, but the main problem is how many OL gain minimum is required to fix all the bad things in the whole amp? Sometimes folded cascode just don't provide enough OL gain to be burned for good CL gain performance.
Other method to linearize OL curve (beside putting R from collector of VAS to ground) is to put 220k in the place of miller cap (base-collector of VAS) and this works good too 😀 But my experience, load to ground or 220k in VAS local loop does not affect stability much, the best way to stabilize an amp is still put caps.
Hi, PPL,
Where's the folded casocode? Connection of Q3-Q4 to Q5-Q6 seems to be double differential? This one has huge OL gain, folded cacode has very low gain
Where's the folded casocode? Connection of Q3-Q4 to Q5-Q6 seems to be double differential? This one has huge OL gain, folded cacode has very low gain
Hi Ppl,
I am looking for the folded cascode, but I cannot see it.
Have you posted the correct schematic?
I am looking for the folded cascode, but I cannot see it.
Have you posted the correct schematic?
Well if taken from a academic point of view the addition of Q5 and Q6 changed the topology from the folded Cascode to a Jfet/BJT Cascode Input Diff amp cascaded to a BJT Cascode Diff amp loaded by a Current Mirror, however that’s too long of a file name after this design is stable without compensation capacitors I can do a PCB up and its then fit to release to the DIY headphone community the name can then be changed to something totally non technical. Attached is what this project started out as and as you can see it is technically a folded Cascode, it actually works for such a simple circuit but not real stellar THD and noise. After looking real close at the schematic to the OPA627/OPA637 IC monolithic Op-Amp’s on the cover of it’s data sheet I saw the pathway to lots of OLG. http://focus.ti.com/lit/ds/symlink/opa627.pdf
Attachments
Folded Cascode Circuit
Hi ppl,
Actually I had very good results with it sonically. You can improve on the current mirror. Cascode the FETs. See my KISS amp:
http://www.diyaudio.com/forums/showthread.php?postid=507432#post507432
You current circuit similar to the OPA627 schematic I started with....
😎
ppl said:
Hi ppl,
Actually I had very good results with it sonically. You can improve on the current mirror. Cascode the FETs. See my KISS amp:
http://www.diyaudio.com/forums/showthread.php?postid=507432#post507432
You current circuit similar to the OPA627 schematic I started with....
😎
Re: Folded Cascode Circuit
Thanks alot for the KISS amp link I must say you used about every Trick in the Book on the Front end of that. I myself however like to use LED voltage references as you also do, I i remember Right Green is low noise and capacitence. So if i used some of the technique you used in your Kiss Amp in my Now apperently incorrectly named Headphone Amp, this would be ok with you? I really like the second gain stage current mirror. A three transistor mirror. and refering the reference back to. Also like yourself it was the OPA627 that inspired my topology also. Strange how manny different people will be thinking the same thing all at the same time, that is so Cool!Elso Kwak said:
Hi ppl,
Am I seeing double on 1st Easter day as I see a 5-transistor current mirror?
Of course you can use it. I found it in a book: Electronische Implementatiekunde by Gerard C. M. Meyer, Delftse Universitaire Pers. (page 138). I only replaced the diode-connected transistor by a green LED. You will find it also in ICs where transistors are used at abundance. Old Harris data-books are full of them. It's a cascoded current mirror to counteract the Early effect
Which reference are you referring to?
😎
Am I seeing double on 1st Easter day as I see a 5-transistor current mirror?
Of course you can use it. I found it in a book: Electronische Implementatiekunde by Gerard C. M. Meyer, Delftse Universitaire Pers. (page 138). I only replaced the diode-connected transistor by a green LED. You will find it also in ICs where transistors are used at abundance. Old Harris data-books are full of them. It's a cascoded current mirror to counteract the Early effect
Which reference are you referring to?
😎
Hi, Elso Kwak,
In the KISS schematic, the differential is CCS'ed by BC550C running at 80? This BC550C works OK with 80V?
In the KISS schematic, the differential is CCS'ed by BC550C running at 80? This BC550C works OK with 80V?
Fatal Errors
Hi lumanauw,
There are some fatal errors in the schematic!
You spotted one of them.
lumanauw said:
Hi lumanauw,
There are some fatal errors in the schematic!
You spotted one of them.
Thanks I will put that into my last version of my Design and see what it sim’s like afterwards. The Three transistors Wilson type current mirror I was referring to are the upper MAT-02 with the BC5460? I did not refer to the lower Cascode part of that Yes 5 transistor Cascode Current mirror. What Harris op amp uses that topology? The reference I was referring to is the Green LED between the bases of the upper and lower Mat-02’s
Oh BTW so has anyone come to a conclusion if a Complementary Differential diff amp stage is an improvement upon these Circuits like in the KISS Amp.
On another note the U440 is a great dual fet I have used them a lot. I don’t like the sound of the Mat series of Duals from ADI, These impart an IC op Amp quality to the sound. The Best sounding Duals both BJT and fet are from these guys http://www.linearsystems.com/ They sell direct and have superb electrical matching. But what attracts me to this company is the sound of there devices, for example the U-440 is available from both LIS and Fairchild yet the LSI part has lower DC offset and like the Fairchild U440 dose not sound like an Op amp, more over is you like the u-44 you will love the Blackness in the background that the LS840 and LS843 give, you do lose some crispness on the top octave with the LS parts over the U440 but micro dynamics to Die for with a properly biased LS-840,LS-843
LSI data sheet U-440 http://www.linearsystems.com/datasheets/U440.pdf
LSI data sheet LS840 http://www.linearsystems.com/datasheets/LS840-2.pdf
LSI data sheet LS843 http://www.linearsystems.com/datasheets/LS843-5.pdf
For Dual Complementary BJT’s I like the LS312 and LS352 better than the ADI parts however if theADI device is all that’s available ill use them.
Oh BTW so has anyone come to a conclusion if a Complementary Differential diff amp stage is an improvement upon these Circuits like in the KISS Amp.
On another note the U440 is a great dual fet I have used them a lot. I don’t like the sound of the Mat series of Duals from ADI, These impart an IC op Amp quality to the sound. The Best sounding Duals both BJT and fet are from these guys http://www.linearsystems.com/ They sell direct and have superb electrical matching. But what attracts me to this company is the sound of there devices, for example the U-440 is available from both LIS and Fairchild yet the LSI part has lower DC offset and like the Fairchild U440 dose not sound like an Op amp, more over is you like the u-44 you will love the Blackness in the background that the LS840 and LS843 give, you do lose some crispness on the top octave with the LS parts over the U440 but micro dynamics to Die for with a properly biased LS-840,LS-843
LSI data sheet U-440 http://www.linearsystems.com/datasheets/U440.pdf
LSI data sheet LS840 http://www.linearsystems.com/datasheets/LS840-2.pdf
LSI data sheet LS843 http://www.linearsystems.com/datasheets/LS843-5.pdf
For Dual Complementary BJT’s I like the LS312 and LS352 better than the ADI parts however if theADI device is all that’s available ill use them.
- Status
- Not open for further replies.