Did you reply to my PM , I want to mail this tomorrow. I am going to make the PB60 overviews , print them to include with the package. Will also update the AX BX documentation.
Speaking of headphones , I listen for hiss, hum , other inconsistencies with 220R resistors hooked to PB250 OPS. I will feel safer doing my "gruntwork" on the 60's.
OS
Speaking of headphones , I listen for hiss, hum , other inconsistencies with 220R resistors hooked to PB250 OPS. I will feel safer doing my "gruntwork" on the 60's.
OS
mongrel amp pictures by repeet_photos - Photobucket
Heres some pics of my progress so far with my Mongrel's.
Just waiting on the semi-conductors to arrive to finish the boards. I also need to make some inductors to finish the PSU's.
Heres some pics of my progress so far with my Mongrel's.
Just waiting on the semi-conductors to arrive to finish the boards. I also need to make some inductors to finish the PSU's.
They look real cool
, man. VERY good work! I will post my new ones , which ,by the way you can attach to your PB120 .. as well. OS
Thanks OS, I haven't got much of a clue about how they actually work, but can follow instructions pretty well. Thanks for all your hard work in developing this design, its been fun so far learning how to make the pcb's as well as having to go through the process of selecting all the components instead of just buying a kit with everything supplied. I luckily had the big gold Obligato caps in my parts box that happened to be 4.7uF so had to use them. I think they look cool.
Pete
Pete
Member
Joined 2009
Paid Member
The thread starter REBORN -CX1.2 BJT and FET
Been a workin' on the 2 best designs I have run into lately. That is , designs that are affordable and approach PPM levels (almost - hundreds???). I must thank Nico Ras .. his "ELD" and the two crazy "goldmund" threads for "triggering" this.
Now , down to business. My goals for the CX's were to :
#1 - have 60+V symmetrical slew rate ,#2 - Have a high bandwidth as well (200k+ / -1db) , #3 - use the goldmund front end but not have it cost an arm and a leg , and finally , my favorite .. #4 - THD as low as it will go... plus dominant H2 for what distortion is left.
What I have come up with to address these design needs is relatively simple. (Picture 1 and 2 are the CX1.2BJT / FET).
I was going to use the current mirror AND the beta enhanced VAS but could not degenerate the stages enough to keep my loop gain under 125DB !!!
The "core" amp, (the CX/ BJT) works with a typical input differential, heavily degenerated,feeding the beta enhanced balanced VAS which uses a high gain full wilson current mirror at it's "bottom end".
Even without the input stage current mirror I get 300ppm THD / 1k , 80db loop gain and a sweet -110db - -125+db FFT (very low high order harmonics)
(PICTURES 3/4 bode and FFT) All in all ... a VERY "transparent" amp. PS - I did run the FFT with a single pair BJT NJW0281/0302 pair @ 75mA !! Still VERY good.
We love the very fast slew , with a 4 V square wave input I could swing the whole 150 volts of my rails in 1.2uS ! (120v/us) This was with 33pF miller compensation. The amp still has over 85 degrees margin at this speed , so it would most likely be very stable ( same as my slower GX). (PICTURES 5/6) are of the slew (about 70V/uS with 47pF millers) , and the bandwidth while leaving the input filter in place ... it would be higher.
COST $$$$$$$$ .... this amp, parts wise... is very low cost .The J201 Jfet is currently $0.41usd at mouser , the 1845/992's are a hefty $0.20/0.08 usd and the 1381/3503's are a WHOPPING $0.29 apiece. In otherwords , a pair of CX's can be made for a whole $20... oohhhhh!
Before I forget , the CX voltage boards will operate with 40-80v rails and can be used with BJT , V mosfet , or lateral mosfets. If somebody could help me design a lateral board with split supplies , a goldmund "killer" could be quite easy to pull off.
Been a workin' on the 2 best designs I have run into lately. That is , designs that are affordable and approach PPM levels (almost - hundreds???). I must thank Nico Ras .. his "ELD" and the two crazy "goldmund" threads for "triggering" this.
Now , down to business. My goals for the CX's were to :
#1 - have 60+V symmetrical slew rate ,#2 - Have a high bandwidth as well (200k+ / -1db) , #3 - use the goldmund front end but not have it cost an arm and a leg , and finally , my favorite .. #4 - THD as low as it will go... plus dominant H2 for what distortion is left.
What I have come up with to address these design needs is relatively simple. (Picture 1 and 2 are the CX1.2BJT / FET).
I was going to use the current mirror AND the beta enhanced VAS but could not degenerate the stages enough to keep my loop gain under 125DB !!!
The "core" amp, (the CX/ BJT) works with a typical input differential, heavily degenerated,feeding the beta enhanced balanced VAS which uses a high gain full wilson current mirror at it's "bottom end".
Even without the input stage current mirror I get 300ppm THD / 1k , 80db loop gain and a sweet -110db - -125+db FFT (very low high order harmonics)
(PICTURES 3/4 bode and FFT) All in all ... a VERY "transparent" amp.
PS - I did run the FFT with a single pair BJT NJW0281/0302 pair @ 75mA !! Still VERY good.We love the very fast slew , with a 4 V square wave input I could swing the whole 150 volts of my rails in 1.2uS ! (120v/us) This was with 33pF miller compensation. The amp still has over 85 degrees margin at this speed , so it would most likely be very stable ( same as my slower GX). (PICTURES 5/6) are of the slew (about 70V/uS with 47pF millers) , and the bandwidth while leaving the input filter in place ... it would be higher.
COST $$$$$$$$ .... this amp, parts wise... is very low cost .The J201 Jfet is currently $0.41usd at mouser , the 1845/992's are a hefty $0.20/0.08 usd and the 1381/3503's are a WHOPPING $0.29 apiece. In otherwords , a pair of CX's can be made for a whole $20... oohhhhh!
Before I forget , the CX voltage boards will operate with 40-80v rails and can be used with BJT , V mosfet , or lateral mosfets. If somebody could help me design a lateral board with split supplies , a goldmund "killer" could be quite easy to pull off.
Attachments
Hey , Bigun. I've done that ,too.
Let me show you.....
(Picture 1 ) shows the CX output (.0001%) with it's cool -120db+ harmonics and how the OPS lacking any global feedback bring this up to .03% with a lot more H3.
(picture 2) is the actual comparison (overlay) of the 2 outputs (drivers / OPS).
Now , the genesis 50W class A does not do this and has just .0004% distortion with no feedback. Laterals are also more benign in their odd harmonic "contributions" to the mix. I would lean in their direction for a no GNF output stage.
OS
Attachments
Hi Pete,
Your Mongrel boards look very nice. I like your way of printing the silkscreen on top of the PB120, which help the parts stuffing.
Have a look of post 695, I used dowel as inductor former and I can wind tight two layers coil and the size seems to fit your HackerCap boards. BTW, you may need small heatsinks for the TO-220 diodes if you use the amp for PA or parties.
It seemed that a couple of resistors are too big for the PB120 but I like the Dale CCF60 & CN55 resistors - they are non-magnetic.
Cheers, Stanley
Attachments
Last edited:
Member
Joined 2009
Paid Member
Hey , Bigun. I've done that ,too.
Let me show you.....
(Picture 1 ) shows the CX output (.0001%) with it's cool -120db+ harmonics and how the OPS lacking any global feedback bring this up to .03% with a lot more H3.
(picture 2) is the actual comparison (overlay) of the 2 outputs (drivers / OPS).
Now , the genesis 50W class A does not do this and has just .0004% distortion with no feedback. Laterals are also more benign in their odd harmonic "contributions" to the mix. I would lean in their direction for a no GNF output stage.
OS
I don't want to operate ClassA.
Even the CX without global feedback looks pretty good at only 0.03%. How readily do you think this would be achievable ? The front end can address the relative balance between H2/H3 for those who care about such things.
Why do you feel BJTs are worse than MOSFETs - is this beta droop that drives H3 ?
you got it , the lateral's large capacitance has a different interaction on the driver pair than the pure transconductance of the NJW's. I have tried the IRFP240/9240 on these creations , much less H3.
Achievable ???... already there , man. I listen to the GX/(or CX) without the OPS through headphones , breathtaking!! the NJW's WILL do .02-.05% without feedback , no question there. That is why they are so easy to to coax to .0005% or below in a GNFB setup.
This is why I need a MOSFET "testbed".
The CX should be flawless , almost no THD (ppm) in isolation("flirts" with the noise floor). a perfect driver/voltage stage for a ppm class A power stage or low THD AB creation.
OS
Last edited:
The IRFP240/9240 are "hardcore" V mosfets , bias them at 200ma ... they don't care.
I did take a pair of the original genesis IRFP's and power them with the original "supersym" voltage stage.... worked well!! (below) pay attention to driver Re (220r - R40) and Vbe ratio (10k/680-145R ..R36-38).
I got 350 PPM @ 1k 40v p-p with this one.!!
OS
Attachments
32's and 33's are quite "beefy" . Quasi (member) drives 5 pair (NMOS 350/500) of the IRF's with paltry MJE340/50's. Sky is the limit. might want to lower the driver Re from 220 to 100R (R40) to negotiate the huge gate capacitance of 5 or more pairs of the IRF's. BTW , you would have a 1/2 KW amp at this point w/80v rails!!
OS
Show an example , however simple it may be.
Is it similar to the "rush" ??? check PM , box is sent today (yesterday - with mica's)
OS