I switched to N P N P layout in 2016.I saw that in a Perreaux amp in the 80's. Not saying they were first but it's at least a 40yo innovation
How about using matched version NPN/NPN or PNP/PNP in one case?
BCM846, 847
BCM856, 857
https://www.mouser.sk/c/?q=bcm846
https://www.mouser.sk/c/?q=bcm847
https://www.mouser.sk/c/?q=bcm856
https://www.mouser.sk/c/?q=bcm857
BCM846, 847
BCM856, 857
https://www.mouser.sk/c/?q=bcm846
https://www.mouser.sk/c/?q=bcm847
https://www.mouser.sk/c/?q=bcm856
https://www.mouser.sk/c/?q=bcm857
The Nexperia dual devices PMP4201 / PMP5201 and PMP4501/PMP5501 also look interesting.
PMP4201/PMP5201 are matched to 98% and the PMP4501/PMP5501 are matched to 95%. Being contained in the same package would provide better thermal contact than any through hole pair will attain.
They're spec'ed at 175MHz with good noise figures (3.3dB or less). Gain is good at 250, but through hole devices are better.
They're tiny though... SOT363 (SC-88) is the biggest package.
PMP4201/PMP5201 are matched to 98% and the PMP4501/PMP5501 are matched to 95%. Being contained in the same package would provide better thermal contact than any through hole pair will attain.
They're spec'ed at 175MHz with good noise figures (3.3dB or less). Gain is good at 250, but through hole devices are better.
They're tiny though... SOT363 (SC-88) is the biggest package.
What makes SM tricky with class A and class AB as opposed to class D and digital amplifiers is heat and the fact that the heatsink already eats up quite some real estate. However having a modular swappable input stage does sound exciting. Complex low current stages can be done in SMT if need be like some well known manufacturers do for discrete opamp sections. Good job.
I just forgot to explain why I did it that way. According to Cherry the high current loop should be as small as possible and by that layout configuration those loops are smaller.
Yes , Me and Keentoken discussed cherry's current loops , also went into the loop of the NFB grounding.
After the Wolverine team playing around with with these ground loop routings , I now see the optimal way.
OST mini will definitely support sub ppm operation.
Now on the the SMD stuff. SOT-363 is just too freaking small !!! The 363 is just one package (with .6mm LS).
.6mm gives me reservations concerning creepage !
There are bigger TSSOP-6 / TSOP-6 / smaller SOT-563/SOT26-6. All have 1mm lead spacing.
P - N pairs have a different pinout than N or P pairs. Checked ZETEX/ON/ diodes inc. ... all the same. (below).
So I came up with a 2 universal pads for all the SOT/SOP's. 90% of all Mouser 6-pin packages (NO 363's - too small !!).
Zetex actually has the same pinout for the matched P - N's.
OS
After the Wolverine team playing around with with these ground loop routings , I now see the optimal way.
OST mini will definitely support sub ppm operation.
Now on the the SMD stuff. SOT-363 is just too freaking small !!! The 363 is just one package (with .6mm LS).
.6mm gives me reservations concerning creepage !
There are bigger TSSOP-6 / TSOP-6 / smaller SOT-563/SOT26-6. All have 1mm lead spacing.
P - N pairs have a different pinout than N or P pairs. Checked ZETEX/ON/ diodes inc. ... all the same. (below).
So I came up with a 2 universal pads for all the SOT/SOP's. 90% of all Mouser 6-pin packages (NO 363's - too small !!).
Zetex actually has the same pinout for the matched P - N's.
OS
Attachments
Here is one to wrap a brain around ....
which combo's of P-N pairs would benefit most from the matched monolithic packages ??
- Q3-7 / Q4-8 or Q3-4 / Q7-8 ?
A quad matched package would be best , but are rare and bloody expensive $$. These quad packages only
match the N or P's , not N to P.
With this circuit , It might even be easier to just to use singles. The Spooky is both dual cascodes or LTP pairs of the same gender - where the
matching is better.
OS
which combo's of P-N pairs would benefit most from the matched monolithic packages ??
- Q3-7 / Q4-8 or Q3-4 / Q7-8 ?
A quad matched package would be best , but are rare and bloody expensive $$. These quad packages only
match the N or P's , not N to P.
With this circuit , It might even be easier to just to use singles. The Spooky is both dual cascodes or LTP pairs of the same gender - where the
matching is better.
OS
Attachments
Would be the same as any other (type) of thermal cycling , if you do not exceed device limits.
Wahhh ! had to go bigger (50 X 70mm) - below. See C3 /C4 , they equal - https://www.mouser.com/datasheet/2/212/KEM_A4070_A765-1101579.pdf
Nothing but the best for my regulators.
Just the input cap and 2 (1uF) servo caps - https://www.mouser.com/datasheet/2/315/ABD0000C178-947606.pdf Price rises to $ 5.00 for 2.2uF !
Input cap will be one of 2 TH components - just a Wima poly 4.7uF. Input terminal will be a 2-position mini-euro.
Spook uses the same VAS/Servo ... so I steal these from this one.
OS
Nothing but the best for my regulators.
Just the input cap and 2 (1uF) servo caps - https://www.mouser.com/datasheet/2/315/ABD0000C178-947606.pdf Price rises to $ 5.00 for 2.2uF !
Input cap will be one of 2 TH components - just a Wima poly 4.7uF. Input terminal will be a 2-position mini-euro.
Spook uses the same VAS/Servo ... so I steal these from this one.
OS
Attachments
Strange combination
, request from my friend, he wants to use kiwane resistors (this one is just protype to test the funtion).Yes with paste.
Dadod , do you ground return your CFA NFB to the speaker ground or the decoupling ground.
The Wolverine team played around with returns , and were able to squeeze lower PPM THD with just slight routings.
(below ) I allowed for either on the bottom (grounding) side of the board - either G2 IPS star OR speaker ground (G1).
CFA will return 50+ mA at full power. I also attenuated the NFB through R5 - 1Mohm to feed the servo .... clean !!
I might just standardize and make these IPS's 76 x 50mm , spook has a few more parts. Still smaller than a credit card - 90 X 55mm.
What I meant - does the input/servo ground referencing off of the (very dirty) NFB ground matter ?
OS
The Wolverine team played around with returns , and were able to squeeze lower PPM THD with just slight routings.
(below ) I allowed for either on the bottom (grounding) side of the board - either G2 IPS star OR speaker ground (G1).
CFA will return 50+ mA at full power. I also attenuated the NFB through R5 - 1Mohm to feed the servo .... clean !!
I might just standardize and make these IPS's 76 x 50mm , spook has a few more parts. Still smaller than a credit card - 90 X 55mm.
What I meant - does the input/servo ground referencing off of the (very dirty) NFB ground matter ?
OS
Attachments
Last edited: