Hi Ostripper
A while ago someone asked about the purpose of the 68pF capacitor which is labeled "C8 (optional)" in the latest schematic.
As far as I can see, it plays an important role in improving power supply rejection.
When the positive supply voltage is jumping up and down wrt earth, C9 injects junk current into the base of Q7. C8 compensates for this by injecting a similar junk current into the base of Q8.
Similarly, the junk current flowing out the bottom of C8 into the emitter of Q9 goes round the mirror to the VAS output, where it cancels out the junk current flowing out the bottom of C9 to the VAS output.
C8 and C9 have to be the same value for this to work but even if you have a 10% mismatch in their values due to component tolerance, you still get a 20dB improvement in PSRR.
Around that time, a couple of variations were proposed, most of which would have made PSRR worse. Eg: if the bottom of C8 is connected to the collector of Q10 instead of the emitter of Q9, PSRR of the positive rail will still be good, but PSRR of the negative rail will be screwed up.
I'm glad to see you stuck to the original.
Sorry about chipping in with this so late but my tatty old grey matter seems to spit out useful insights somewhat randomly as and when it feels like it, rather than when required.
Cheers - Godfrey
A while ago someone asked about the purpose of the 68pF capacitor which is labeled "C8 (optional)" in the latest schematic.
As far as I can see, it plays an important role in improving power supply rejection.
When the positive supply voltage is jumping up and down wrt earth, C9 injects junk current into the base of Q7. C8 compensates for this by injecting a similar junk current into the base of Q8.
Similarly, the junk current flowing out the bottom of C8 into the emitter of Q9 goes round the mirror to the VAS output, where it cancels out the junk current flowing out the bottom of C9 to the VAS output.
C8 and C9 have to be the same value for this to work but even if you have a 10% mismatch in their values due to component tolerance, you still get a 20dB improvement in PSRR.
Around that time, a couple of variations were proposed, most of which would have made PSRR worse. Eg: if the bottom of C8 is connected to the collector of Q10 instead of the emitter of Q9, PSRR of the positive rail will still be good, but PSRR of the negative rail will be screwed up.
I'm glad to see you stuck to the original.
Sorry about chipping in with this so late but my tatty old grey matter seems to spit out useful insights somewhat randomly as and when it feels like it, rather than when required.
Cheers - Godfrey
Attachments
Hi Ostripper
A while ago someone asked about the purpose of the 68pF capacitor which is labeled "C8 (optional)" in the latest schematic.
As far as I can see, it plays an important role in improving power supply rejection.
When the positive supply voltage is jumping up and down wrt earth, C9 injects junk current into the base of Q7. C8 compensates for this by injecting a similar junk current into the base of Q8.
Similarly, the junk current flowing out the bottom of C8 into the emitter of Q9 goes round the mirror to the VAS output, where it cancels out the junk current flowing out the bottom of C9 to the VAS output.
C8 and C9 have to be the same value for this to work but even if you have a 10% mismatch in their values due to component tolerance, you still get a 20dB improvement in PSRR.
Around that time, a couple of variations were proposed, most of which would have made PSRR worse. Eg: if the bottom of C8 is connected to the collector of Q10 instead of the emitter of Q9, PSRR of the positive rail will still be good, but PSRR of the negative rail will be screwed up.
I'm glad to see you stuck to the original.
Sorry about chipping in with this so late but my tatty old grey matter seems to spit out useful insights somewhat randomly as and when it feels like it, rather than when required.
Cheers - Godfrey
Thanx for input , I have not explored that aspect of C8 but I left the pads for it , just in case. What I did for PSRR is added the capacitance multipliers and referenced both the cascode and CCS to the rails . On the simulation I added 4V p-p @ 120 hz AC to the rails and measured PSRR ... --80DB + all across the audio band. On the amp in the picture I have driven full house shaking power and heard NOTHING on the undriven channel .... I tried this with and without the second 68pF cap.
As far as sticking to the original , "If it ain't broke , tweak it more" . I like the originals sound and it is 13 months old , no
This one is just a pro replacement for it.My bench .. HA ! it is my desk .. I am poor. (see below) I was lucky to get my 20mhz scope and a amp. As far as funds , I have to wait till august to print the boards (plenty of time to check for errors. BTW , I have just burnt 1
njw0302 in 2 years and a dozen amps.
OS
Attachments
"When the positive supply voltage is jumping up and down wrt earth, C9 injects junk current into the base of Q7. C8 compensates for this by injecting a similar junk current into the base of Q8.
Similarly, the junk current flowing out the bottom of C8 into the emitter of Q9 goes round the mirror to the VAS output, where it cancels out the junk current flowing out the bottom of C9 to the VAS output."
The collector of Q8 is held at 1 diode drop above 0V by Q9. If you look at the base of Q8, it will only be moving around by a very small amount. On the Q9 side, C9 has very large voltage swings across it because one end is tied to the LTP collector load and the other to the VAs output, so the situation is completely different, and substantial capacitor currents therefore flow in C9, compared to C8.
I don't think C8 does anything here. As far as PSU noise injection goes, a diff pair is fairly immune to rail noise on the collector load (or mirror) side and especially where the diff amp is cascoded as is the case here. Besides, this design has a capacitance multiplier type rail filter - it should be fairly clean.
Similarly, the junk current flowing out the bottom of C8 into the emitter of Q9 goes round the mirror to the VAS output, where it cancels out the junk current flowing out the bottom of C9 to the VAS output."
The collector of Q8 is held at 1 diode drop above 0V by Q9. If you look at the base of Q8, it will only be moving around by a very small amount. On the Q9 side, C9 has very large voltage swings across it because one end is tied to the LTP collector load and the other to the VAs output, so the situation is completely different, and substantial capacitor currents therefore flow in C9, compared to C8.
I don't think C8 does anything here. As far as PSU noise injection goes, a diff pair is fairly immune to rail noise on the collector load (or mirror) side and especially where the diff amp is cascoded as is the case here. Besides, this design has a capacitance multiplier type rail filter - it should be fairly clean.
All PDF's
I converted everything to PDF for smaller DL size.
Here : HFS /WEBSITE/MONGREL/
"overview parts" is a verbose placement map of both boards which coincides with the Mongrel V1.1 schematic. These should look REAL good when printed,etched and stuffed. LTP and cascode are "face to face" for heat-shrink thermal coupling. Power board marked for implementation of EF3 (2 - 3.3k 2w resistors).
To answer the question of "how the boards go together" ,
the voltage amp board will have:
Power board will have :
PS. if you see any errors , please tell ..
Yes , bonsai , I had a hard time verifying this both real world and simulation wise. I did leave the option open on the PCB.
OS
I converted everything to PDF for smaller DL size.
Here : HFS /WEBSITE/MONGREL/
"overview parts" is a verbose placement map of both boards which coincides with the Mongrel V1.1 schematic. These should look REAL good when printed,etched and stuffed.
LTP and cascode are "face to face" for heat-shrink thermal coupling. Power board marked for implementation of EF3 (2 - 3.3k 2w resistors). To answer the question of "how the boards go together" ,
the voltage amp board will have:
Power board will have :
PS. if you see any errors , please tell ..
Yes , bonsai , I had a hard time verifying this both real world and simulation wise. I did leave the option open on the PCB.
OS
Attachments
Last edited:
No more b_s..
Back at it !! This time , NO MORE B-S .... After looking at so many DIY and commercial amps , I think I have what I want , maybe even what others may want.
The previous layout was good and would of even worked well , but it was too specific to my purposes and had a tired, typical look. In order for my DREAM of the universal AB amp to be reality a more useful scheme had to be designed.
I ordered the parts ($171.00 for 6 amps), that is $28 an amp ... not bad.
THE CHANGES. The new board is the PB250 -8 output devices + Dual capacitance multipliers ... capable of 250W 8R /350 - 400W 4R depending on which devices are used. I kept the CX1.1 (symasym voltage amp), since it is perfect.
Here is what will be available. All have actually been built at one time or another (frugalamps - supersyms, etc.)
POWER BOARDS
PB-250 /250W/ biggest BJT board ,70-80V rails..... completed.
PB-70 /70W/ baby BJT board ,power supply built in... all in one!! (work in progress - 8/5/10)
PBM-200 / Vertical mosfet power board (still pondering this one
)
VOLTAGE BOARDS
These are fun ...
AX1.1 - Your typical Doug- Self blameless amp , tweaked to the max ... .0002% at 1khz , .002% at 20khz , definitely the least distortion of them all.
BX1- The bootstrap amp , lousy .02%THD .. but ,when built, very musical indeed!! Mostly 2nd harmonic even as it is the "dirtiest amp".
CX1.1 -balanced "symasym" type amp , seems to have the best characteristics of the AX AND BX , 2nd harmonic dominant but with the super low distortion of the self amp.
NO DX ANYTHING !!
EX - complimentary balanced "leach" type amp (work in progress)
FX - Harmon Kardon "signature series" amp - very promising (also a work in progress).
One might think "why is he doing this ??" For one , I am selling these .. so I make my $$$'s back. Also , I learned just enough to be dangerous here at the forum , so I must recontribute .
Lastly , I am real disturbed by the direction the forum is taking. After reading though the DX BLAME ES and ST threads .. OH MY GOD !! The ego is through the roof , the engineering is buried beneath the floor. The AX1.1 would TOTALLY outperform the ST , better sound stage and much more stable (will run without grounded heatsinks, zoble, or a BIG STUPID 100PF miller cap !! ).
Below is examples of New board layout and the AX1.1 (I am building a pair of these, too).
ALL is available at : HFS /WEBSITE/MONGREL/
OS
Back at it !! This time , NO MORE B-S .... After looking at so many DIY and commercial amps , I think I have what I want , maybe even what others may want.
The previous layout was good and would of even worked well , but it was too specific to my purposes and had a tired, typical look. In order for my DREAM of the universal AB amp to be reality a more useful scheme had to be designed.
I ordered the parts ($171.00 for 6 amps), that is $28 an amp ... not bad.
THE CHANGES. The new board is the PB250 -8 output devices + Dual capacitance multipliers ... capable of 250W 8R /350 - 400W 4R depending on which devices are used. I kept the CX1.1 (symasym voltage amp), since it is perfect.
Here is what will be available. All have actually been built at one time or another (frugalamps - supersyms, etc.)
POWER BOARDS
PB-250 /250W/ biggest BJT board ,70-80V rails..... completed.
PB-70 /70W/ baby BJT board ,power supply built in... all in one!! (work in progress - 8/5/10)
PBM-200 / Vertical mosfet power board (still pondering this one
)VOLTAGE BOARDS
These are fun ...
AX1.1 - Your typical Doug- Self blameless amp , tweaked to the max ... .0002% at 1khz , .002% at 20khz , definitely the least distortion of them all.
BX1- The bootstrap amp , lousy .02%THD .. but ,when built, very musical indeed!!
Mostly 2nd harmonic even as it is the "dirtiest amp".CX1.1 -balanced "symasym" type amp , seems to have the best characteristics of the AX AND BX , 2nd harmonic dominant but with the super low distortion of the self amp.
NO DX ANYTHING !!
EX - complimentary balanced "leach" type amp (work in progress)
FX - Harmon Kardon "signature series" amp - very promising (also a work in progress).
One might think "why is he doing this ??" For one , I am selling these .. so I make my $$$'s back. Also , I learned just enough to be dangerous here at the forum , so I must recontribute .
Lastly , I am real disturbed by the direction the forum is taking. After reading though the DX BLAME ES and ST threads .. OH MY GOD !!
The ego is through the roof , the engineering is buried beneath the floor. The AX1.1 would TOTALLY outperform the ST , better sound stage and much more stable (will run without grounded heatsinks, zoble, or a BIG STUPID 100PF miller cap !! ).Below is examples of New board layout and the AX1.1 (I am building a pair of these, too).
ALL is available at : HFS /WEBSITE/MONGREL/
OS
Attachments
Yes and Yes.
All will be the same standard.
I had to wrangle how to do the parts identification with a modular system. With most combo's I number the power boards R99,R98 / C99,C98 backwards. All voltage boards go up R1,R2... No conflicts any combination.
I am not sure about selling boards unless there is interest. Now that I know how to make them I might keep a couple handy for the mailbox.
Next is the AX1 and the PB70 , might have them done tomorrow.
By friday the major 3 amps AX,BX,CX and the 2 power boards will be submitted.Thank you , Kean..
OS
Last edited:
I have the "little mongrel" done , the "PB60" (below - 1). Stuffed with enough capacitance (2 X 6800 - 10000uF / 30mm snap-in devices) to be a stand alone board ... just add a bridge , this little board will take all the X voltage boards (BX is next).
I LOVE MY NEW $20 SCOPE !!
I have been scoping traces on my Supersym and noticed that ripple increases drastically just centimeters away from the star ground. This influenced my design criteria for the new boards. The current pulses in the rails cancel out completely at a common center point . This I saw in a minor way with my small 60W amps but to a larger degree on my huge amp. THE LAYOUT IS IMPORTANT , visually confirmed...
Another observation is with the current balance in all my LTP's .... just a small adjustment in the current source (.1 - .2mA) makes all the difference in the world for the soundstage in ANY amp. Now all 4 of my DIY amps have the adjustable CCS. I truly wonder why this has not been common practice with amps on the forum. In simulation and backed up on my scope , sound quality and performance are worthy of this tweak , as well as the ability to adjust for different power supply rails. An adjustment range of 1 - 2 mA seems to cover all devices and rail voltages.
Below are the plots for the AX , BX , and CX (supersym 1) differential currents (Q1 and 2) ...with perfect balance the distortion is halved , mostly higher order stuff. This can be HEARD in real world , too !! The self amp ,the "AX" , is extremely good , but something about a current sourced VAS just doesn't sound as good. BUT ..I will try them again. BTW , the CX is doing 125V p-p at 20khz.
After seeing what these simple circuits can really do , I am going to build all 3
to swap them out on my PB250 board. They only cost about $ 10 apiece .. board and all.
OS
PS , The artwork for the CX1.1VB and PB250 (MY NEW AMP) are totally done and quadruple checked ,the most common mouser parts will fit perfectly and the BOM is done.
(3rd pix) How they go together.. he he
I LOVE MY NEW $20 SCOPE !!
I have been scoping traces on my Supersym and noticed that ripple increases drastically just centimeters away from the star ground. This influenced my design criteria for the new boards. The current pulses in the rails cancel out completely at a common center point . This I saw in a minor way with my small 60W amps but to a larger degree on my huge amp. THE LAYOUT IS IMPORTANT , visually confirmed...
Another observation is with the current balance in all my LTP's .... just a small adjustment in the current source (.1 - .2mA) makes all the difference in the world for the soundstage in ANY amp. Now all 4 of my DIY amps have the adjustable CCS. I truly wonder why this has not been common practice with amps on the forum. In simulation and backed up on my scope , sound quality and performance are worthy of this tweak , as well as the ability to adjust for different power supply rails. An adjustment range of 1 - 2 mA seems to cover all devices and rail voltages.
Below are the plots for the AX , BX , and CX (supersym 1) differential currents (Q1 and 2) ...with perfect balance the distortion is halved , mostly higher order stuff. This can be HEARD in real world , too !!
The self amp ,the "AX" , is extremely good , but something about a current sourced VAS just doesn't sound as good. BUT ..I will try them again. BTW , the CX is doing 125V p-p at 20khz. After seeing what these simple circuits can really do , I am going to build all 3
to swap them out on my PB250 board. They only cost about $ 10 apiece .. board and all.
OS
PS , The artwork for the CX1.1VB and PB250 (MY NEW AMP) are totally done and quadruple checked ,the most common mouser parts will fit perfectly and the BOM is done.
(3rd pix) How they go together.. he he
Attachments
Thanks for posting guys!! I won't bug "uncle charlie" anymore as this will give me time to finish all of these !!
I want these amps to be based on MATH and theory , as well as on the final listening experience. The AX , BX , CX and both powerboards PB60/250 ,I submit with the utmost confidence since they have already been built and simulated to "death" in the "mongrel labs".
The PB60/ 250 are totally finished.They have been reviewed/printed and triple checked for accuracy. You can download these in the ROOT of the mongrel folder as "PB60_schematic_artwork_Bom_complete.ZIP" and "PB250_schematic_artwork_BOM complete.ZIP".
They are smaller at about 1-2 megs but contain the main PCB 1200DPI artwork , parts placement , mirrored front board screenprint and detailed schematic and BOM. This took me a little longer to do because I wanted to perfect a "system" to facilitate further expansion. By the way the "CX1.1_ _schematic_artwork_BOM_complete.ZIP" is also in the "DONE" pile.
The voltage boards will accommodate ANY voltage with a simple adjustment of the "CCS ADJ." trimmer provided, if you stuff the board with the 80/100V capacitors that are recommended. All the semiconductors are the High voltage types (120-300v) - Not BCxxx's , well over-rated for anything. The BC60 is more suited for 40-50V rails as It is just a "little amp" with 63V caps (smaller nichicon 22 mm w/10mm lead spacing or ALL the snap-ins which also have 10mm spacings).
The larger board (PB250)will do 80V rails provided you do use 100v electrolytic's , with 80v units 60-70V rails are recommended.
Thanks , KEAN . If the don't like the CCS's (I heard little difference) they can do the BX (which will be my tribute to the ASKA). It will have the pads to do all sorts of mystical compensations or to leave the CCS out altogether. combine the BX with the PB250 and you will have an ASKA55 that will do 400w peaks!!
I do have the grounding "licked". By scoping real amps, what I have done on the powerboards is "state of the art " , giving almost ZERO hum and /or ripple at the 2 center nodes of the star(s). With the Capacitance multipliers on the 250 , the voltage boards "think" they have 10,000uF decoupling capacitors 47uF X 200Hfe = 10K = uV ripple).
There will be NO PB120 - Instead I will put out a 4 device vertical MOSFET offering (PBM 200).
The ones that are next in the "DONE BIN" are the AX and FX , being almost the same layout .. they are SO different . The AX is the standard blameless with it's .0002% distortion and the FX (KODA super pair).
OS
I want these amps to be based on MATH and theory , as well as on the final listening experience. The AX , BX , CX and both powerboards PB60/250 ,I submit with the utmost confidence since they have already been built and simulated to "death" in the "mongrel labs".
The PB60/ 250 are totally finished.They have been reviewed/printed and triple checked for accuracy. You can download these in the ROOT of the mongrel folder as "PB60_schematic_artwork_Bom_complete.ZIP" and "PB250_schematic_artwork_BOM complete.ZIP".
They are smaller at about 1-2 megs but contain the main PCB 1200DPI artwork , parts placement , mirrored front board screenprint and detailed schematic and BOM. This took me a little longer to do because I wanted to perfect a "system" to facilitate further expansion. By the way the "CX1.1_ _schematic_artwork_BOM_complete.ZIP" is also in the "DONE" pile.
The voltage boards will accommodate ANY voltage with a simple adjustment of the "CCS ADJ." trimmer provided, if you stuff the board with the 80/100V capacitors that are recommended. All the semiconductors are the High voltage types (120-300v) - Not BCxxx's , well over-rated for anything. The BC60 is more suited for 40-50V rails as It is just a "little amp" with 63V caps (smaller nichicon 22 mm w/10mm lead spacing or ALL the snap-ins which also have 10mm spacings).
The larger board (PB250)will do 80V rails provided you do use 100v electrolytic's , with 80v units 60-70V rails are recommended.
Thanks , KEAN .
If the don't like the CCS's (I heard little difference) they can do the BX (which will be my tribute to the ASKA). It will have the pads to do all sorts of mystical compensations or to leave the CCS out altogether. combine the BX with the PB250 and you will have an ASKA55 that will do 400w peaks!!You say elegant enginneering , NO machines are involved here ... all hand drawn , and mental layouts. I would ask you , after looking at a lot of layouts , are the parallel lines in the powerboards (especially the PB250) more prone to parasitics ?? I have seen this style in the quasi amps and many commercial offerings.
I do have the grounding "licked". By scoping real amps, what I have done on the powerboards is "state of the art " , giving almost ZERO hum and /or ripple at the 2 center nodes of the star(s). With the Capacitance multipliers on the 250 , the voltage boards "think" they have 10,000uF decoupling capacitors 47uF X 200Hfe = 10K = uV ripple).
There will be NO PB120 - Instead I will put out a 4 device vertical MOSFET offering (PBM 200).
The ones that are next in the "DONE BIN" are the AX and FX , being almost the same layout .. they are SO different . The AX is the standard blameless with it's .0002% distortion and the FX (KODA super pair).
OS
Attachments
Nice amp!
What´s limiting the slew? Output bipolars? Would MosFets in the output make the amp overall faster?
How much slew do you want ??? The 250 with the CX does 40+ V/uS symmetrical.
The "super pair" (Koda FX1.1 - not done) will do 53+ V/us. That is with a big lunky 100pF miller. (below) I am going to try the FX with the MOSFET board to use as a sub amp. I am still "scared" of the FX , as many have told me it is unstable.
OS
Attachments
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.