I do like the Wolverine .... but it is a little too big and expensive.
Some of us use 2.1 systems with a sub , we don't need 300W/ch. It seems there aren't many small amps that have
super low distortion , either "big and dirty" or small and dirtier ??
-Pre-drivers will directly run the FET gates , full EF3 BJT (NJW0281/0302) with just 3 holes (P / Vbe / N).
- 60 watts for BJT , up to 200W for either MOSFET version. Just a pair of IFRP140/9140 will make a 200W compact plate amp
that will outperform any class AB Dayton setup.
I will make the SMD component choices as large a package as possible. When I order my PCB's I will even offer completed and
tested IPS's.
The output stages will be mostly through-hole except for dual 2W .47R emitter resistors (0.35 each at Mouser !! ) Input stage boards might
just have a through-hole trimmer and the 2 - 12V regulator electro's.
The designs -
SPOOKY (below 2). BEST for BASS !! Tested and revered by slewmaster builders. I might use the Hellraiser servo for it (to eliminate the DC cap).
HELLRAISER (below 3) - also tested in it's raw form back in the Slewmaster days. 250V/us super fast and super low THD CFA. Does not even need
any fancy FB compensation to achieve 2-3ppm performance .
BOTH , can use 39pF SMD silver mica caps (3$ apiece - mouser) still expensive !! 0.30$ high quality ceramics can also be used.
These projects should be far easier even as they are SMD. I will carefully select the largest possible packages to make it easy.
The nice thing , you can just plug these into Stuart's monster Wolverine PCB's. They both have adjustable VAS trimmers.
For the small amp's , I'll run the VAS's at 3.2mA. NO HEATSINKS - except for the output devices/main Vbe !!! Very few large capacitors , as well.
Servo will self adjust down to uV's.
Any other ideas welcome .....
Some of us use 2.1 systems with a sub , we don't need 300W/ch. It seems there aren't many small amps that have
super low distortion , either "big and dirty" or small and dirtier ??
- 50 X 50 mm (wolverine) compatible input stage "blanks" (below 1).
- 60 X 80mm output stage boards.
-Pre-drivers will directly run the FET gates , full EF3 BJT (NJW0281/0302) with just 3 holes (P / Vbe / N).
- 60 watts for BJT , up to 200W for either MOSFET version. Just a pair of IFRP140/9140 will make a 200W compact plate amp
that will outperform any class AB Dayton setup.
I will make the SMD component choices as large a package as possible. When I order my PCB's I will even offer completed and
tested IPS's.
The output stages will be mostly through-hole except for dual 2W .47R emitter resistors (0.35 each at Mouser !! ) Input stage boards might
just have a through-hole trimmer and the 2 - 12V regulator electro's.
The designs -
SPOOKY (below 2). BEST for BASS !! Tested and revered by slewmaster builders. I might use the Hellraiser servo for it (to eliminate the DC cap).
HELLRAISER (below 3) - also tested in it's raw form back in the Slewmaster days. 250V/us super fast and super low THD CFA. Does not even need
any fancy FB compensation to achieve 2-3ppm performance .
BOTH , can use 39pF SMD silver mica caps (3$ apiece - mouser) still expensive !! 0.30$ high quality ceramics can also be used.
These projects should be far easier even as they are SMD. I will carefully select the largest possible packages to make it easy.
The nice thing , you can just plug these into Stuart's monster Wolverine PCB's. They both have adjustable VAS trimmers.
For the small amp's , I'll run the VAS's at 3.2mA. NO HEATSINKS - except for the output devices/main Vbe !!! Very few large capacitors , as well.
Servo will self adjust down to uV's.
Any other ideas welcome .....
Attachments
Last edited:
Yup ! I found the VAS/predriver semi's (in SMD) 2SA1419/2SC3649. https://www.mouser.com/datasheet/2/308/1/EN2007_D-2311156.pdf
160V/ 1.5A/ 120Mhz / 10pF Cob. 0.50$ each in SOT-89. Just 3 sq. CM for the heatsink. 15K in stock. NO sourcing issues with SMD.
LTP and other BCxxx BJT's are easier. Duals are available for everything ... even matched precision pairs. A much shorter build guide.
To streamline I will use these on the OPS , as well. But most of the OPS will be big through-hole with a 14pin SIP at the edge.
Even as these designs cancel rail ripple natively , I'll still include the "ripple eaters" on the OPS. 2SA1419/2SC3649 can be used
for these (1.5A) , as well. So 4 SOT-89 , 2 - TO-220 (drivers) , 2 TO-126 (Vbe's) and the 2 big TO3-P/TO-247 (outputs).
I don't have any artwork for SMD. I'll make a deluxe collection for all the LED's/resistors and semi's. I'll use the datasheets reference.
Sound like good nerdy fun !!
Wow , small ... just 1/2 a CM for all this goodness.
160V/ 1.5A/ 120Mhz / 10pF Cob. 0.50$ each in SOT-89. Just 3 sq. CM for the heatsink. 15K in stock. NO sourcing issues with SMD.
LTP and other BCxxx BJT's are easier. Duals are available for everything ... even matched precision pairs. A much shorter build guide.
To streamline I will use these on the OPS , as well. But most of the OPS will be big through-hole with a 14pin SIP at the edge.
Even as these designs cancel rail ripple natively , I'll still include the "ripple eaters" on the OPS. 2SA1419/2SC3649 can be used
for these (1.5A) , as well. So 4 SOT-89 , 2 - TO-220 (drivers) , 2 TO-126 (Vbe's) and the 2 big TO3-P/TO-247 (outputs).
I don't have any artwork for SMD. I'll make a deluxe collection for all the LED's/resistors and semi's. I'll use the datasheets reference.
Sound like good nerdy fun !!
Wow , small ... just 1/2 a CM for all this goodness.
Attachments
Cool! more SMT options! I'm curious how the Hellraiser sounds? Might be an interesting project to plug into one of Valery's non switching output stages to see how it measures.
Zetex FZT696/FZT795 have been working good for VAS too. https://www.mouser.ca/datasheet/2/115/DIOD_S_A0001070313_1-2542008.pdf
Zetex FZT696/FZT795 have been working good for VAS too. https://www.mouser.ca/datasheet/2/115/DIOD_S_A0001070313_1-2542008.pdf
You may find fitting spooky on a 50mm x 50mm pcb challenging. I started a smd spooky sometime ago (the original version), got most of the pcb layout done then Sanken EOL the driver and op transistors so I then changed the output stage to ECX10N20/P20 lateral mosfets (SpookyFet). The input stage fits in an area 75mm x 62mm, I moved all the DC voltage regulation off board to the main power supply board. VAS transistors are 2sa1552/2sc4027, Accuphase use these in their amplifiers. Source resistors are vishay WSR 0.15R 5watts.
As a suggestion stick everything on a single pcb, it will make laying out the pcb much easier, I'm not a fan of these daughter board arrangements there's no benefit and the pcb area is the same anyway and you now have to purchase two pcbs.
SpookyFet shares its topology from a series of Sherwood lateral mosfet amplifiers S-6040CP, AM-7040 and AM-8500.
As a suggestion stick everything on a single pcb, it will make laying out the pcb much easier, I'm not a fan of these daughter board arrangements there's no benefit and the pcb area is the same anyway and you now have to purchase two pcbs.
SpookyFet shares its topology from a series of Sherwood lateral mosfet amplifiers S-6040CP, AM-7040 and AM-8500.
Attachments
Spooky 2.1 SERVO... had to invert it. opposite of the hellraiser diamond.
Real nice (below). Servo failure = <.5V offset , Bomb proof !
Both amps are servo failure proof and no servo interaction with the FB path - have not seen this advanced trick in any of the Sherwood's.
in fact , the Sherwood's design is CHEAP. FET CSS to set the cascodes ?? Noisy hot Zeners all over ?? I could eliminate my cold zeners with 79L12/78L12 -
-72db ripple rejection IC regulators. Eliminate 6 parts (and 2 zeners).
YES ! Both Digikey and Mouser have 5-10K of each regulator in stock (SOT-89).
Both circuits are final , I can't believe how hard they are to error. And without the regulators (zener/BJT) 100db+ PSRR !
Indiglo ?? Is that DIPTRACE ? Very nice layout. But I want to go manual with sprint so I can familiarize myself with the SMD world.
Customize my pads (curved pads) and really go anal on the layout. Any tips on how to make pads more friendly for either hand or bake
methods. I don't have much SMD experience ??
OS
Real nice (below). Servo failure = <.5V offset , Bomb proof !
Both amps are servo failure proof and no servo interaction with the FB path - have not seen this advanced trick in any of the Sherwood's.
in fact , the Sherwood's design is CHEAP. FET CSS to set the cascodes ?? Noisy hot Zeners all over ?? I could eliminate my cold zeners with 79L12/78L12 -
-72db ripple rejection IC regulators. Eliminate 6 parts (and 2 zeners).
YES ! Both Digikey and Mouser have 5-10K of each regulator in stock (SOT-89).
Both circuits are final , I can't believe how hard they are to error. And without the regulators (zener/BJT) 100db+ PSRR !
Indiglo ?? Is that DIPTRACE ? Very nice layout. But I want to go manual with sprint so I can familiarize myself with the SMD world.
Customize my pads (curved pads) and really go anal on the layout. Any tips on how to make pads more friendly for either hand or bake
methods. I don't have much SMD experience ??
OS
Attachments
Last edited:
Not yet, I still need to route the op stage and decide on connectors and finish the power supply with its overcurrent protection and dc regulators.
My spooky runs vas at 4.5mA, ltspice tells me vas transistor sits around 230mW, I added copper pours for heatsinking.
I did consider this, but I thought about obsolescence, it happens to smd as well, toshiba eol their low noise smd 2sa1312/2sc3324 transistors. This is why I switched to lateral mosfets from the sanken BJT's that went EOL.
Its a good idea, if people want to toaster oven or pan fry their boards they can order a solder paste stencil with the pcb.
I'm using Kicad 7, it has a massive community behind it, I worry about the smaller commercial pcb cad vendors, Kicad will eventually kill them off. My layout was completely manual, I re-laid out the pcb 3 times to get the physical layout looking symmetrical, I found a spooky through hole layout in the old thread that provided ideas to get it where it is now. I spent some time optimising the layout to ensure I didn't have any splits in the bottom ground plane.
I'm not familiar with Diptrace, I used the standard smd pads for each component. I started out with handsolder pads then changed to the standard size after a Kicad update from V6 to V7. The Kicad handsolder resistor pads are only 0.175mm larger than the standard pads.
If people use fine soldering iron tips and solder diameter 0.381mm or less then standard pad sizes will work for handsoldering, toaster oven and pan fried style. It's not possible to cater for everyone otherwise you'll end up with half a dozen board layouts to maintain.
For example take the wolverine input capacitor, it has every conceivable option for every size capacitor taking up about 15% of the pcb area. My view is stick to standard pad sizes it can only make your life easier.
Since you need to cater for the three soldering methods use thermal reliefs for components connecting to a ground plane to prevent tombstoning when using the toaster oven and pan fried pcb solder reflow methods.
Hi OS, that's too bad
!I was looking for a simple IPS (Front End only) board for some OPS that I have around. Do not need a lot of gain, or 0.....1% distorsion, but a lot f VAS current is useful (i do not mind if I have to make some simple mod like a Darlington VAS). I have one of my own design but it is a little bulky and not that good, let's say... so not worthy to design a PCB probably. I would go for something like an LME49810, but
sigh, that is no more available.Also I like small and and SMD.
Is there something in the various iterations of your designs where there are boards available? Sorry, but in these fifteen years here your design production is soo great that my head is spinning searching in the threads...😁
I am at the final stage...
SMD is a learning curve. I had a discussion about whether my servo's were in the audio path. No , they are not. Unless you consider audio induced rail ripple
as audio .... then everything in a design is "in the audio path".
I had to describe my servo as a "voltage controlled current source" , they still said it was a non-linear way to achieve the end result (uV offset / no 10hz added THD). Oh well.
NO problem fitting spooky on a 50 X 50 mm board. Hellraiser will have its 1-2w feedback resistors on the bottom side. I DO have the bottom side on any
SMD design , great for star grounding and "dirty/noisy" parts.
I'll edit my first post for the updated LT simulations. For SMD , I dropped my VAS Ic to 3.7mA. NO added distortion. LATFET or EF3 is nice for such cool low
current IPS's.
NO adjustments , red led of 1.7V is simulated , A newer modern 1.85V red led would just require increasing the current source Re. VAS cascode's LED Vf doesn't matter. NO trimmers , NO big caps in the IPS.
OS
SMD is a learning curve. I had a discussion about whether my servo's were in the audio path. No , they are not. Unless you consider audio induced rail ripple
as audio .... then everything in a design is "in the audio path".
I had to describe my servo as a "voltage controlled current source" , they still said it was a non-linear way to achieve the end result (uV offset / no 10hz added THD). Oh well.
NO problem fitting spooky on a 50 X 50 mm board. Hellraiser will have its 1-2w feedback resistors on the bottom side. I DO have the bottom side on any
SMD design , great for star grounding and "dirty/noisy" parts.
I'll edit my first post for the updated LT simulations. For SMD , I dropped my VAS Ic to 3.7mA. NO added distortion. LATFET or EF3 is nice for such cool low
current IPS's.
NO adjustments , red led of 1.7V is simulated , A newer modern 1.85V red led would just require increasing the current source Re. VAS cascode's LED Vf doesn't matter. NO trimmers , NO big caps in the IPS.
OS
Attachments
Last edited:
I am using IRFPxxx from 2013. (look this thread https://www.diyaudio.com/community/threads/200w-mosfet-cfa-amp.243481/) with Vbe multiplier + red LED and 2SA1381, 2SA3503 as the drivers, and, by the way, not from the slewmaster. U2 is THAT 340.
Damir
Damir
I am progressing. I finally decided on a 4 output baby amp board - 75 X 135mm. Since it's such a small board , I'll recycle the layout to make
a vertical FET version of it.
Same basic WINNING layout as the big Wolverine , it should be a low PPM precision amp. Lots of options , if you are poor .
- Just populate Q109/Q110 if you want a real small cheap amp (60W). Perfect for satellite speakers. Populate all 4 for a 100W amp.
-This layout is small enough to attach to any sub plate amp or E-waste amp chassis.
-You can mount the tiny SMD IPS's at a right angle or flat. A wolverine IPS would also work with it.
I'm also going to make a supply capacitor PCB - 75 X 150mm , with outputs for 2 amps (stereo) and a series regulator + 7812/7912 regulated
aux output. I might use heatsinked TO-220 diodes , snubbers and all the other extra's.
Since this is smaller amp , all caps will be 4700 - 6800uF X4 63V Nichicon's. 30-0-30 ... 40-0-40V transformers are recommended.
Price , with SMD , the IPS's won't cost more that 15$. I'm also using SMD silver mica's (2.65$ apiece).
Output board has a few SMD's , those .1uF decoupling caps are soooo cheap. Easy to just solder them right at the electrolytic leads.
Why buy 1$+ wima .1uF's ?? All the large resistors will be through- hole.
OS
a vertical FET version of it.
Same basic WINNING layout as the big Wolverine , it should be a low PPM precision amp. Lots of options , if you are poor
.- Just populate Q109/Q110 if you want a real small cheap amp (60W). Perfect for satellite speakers. Populate all 4 for a 100W amp.
-This layout is small enough to attach to any sub plate amp or E-waste amp chassis.
-You can mount the tiny SMD IPS's at a right angle or flat. A wolverine IPS would also work with it.
I'm also going to make a supply capacitor PCB - 75 X 150mm , with outputs for 2 amps (stereo) and a series regulator + 7812/7912 regulated
aux output. I might use heatsinked TO-220 diodes , snubbers and all the other extra's.
Since this is smaller amp , all caps will be 4700 - 6800uF X4 63V Nichicon's. 30-0-30 ... 40-0-40V transformers are recommended.
Price , with SMD , the IPS's won't cost more that 15$. I'm also using SMD silver mica's (2.65$ apiece).
Output board has a few SMD's , those .1uF decoupling caps are soooo cheap. Easy to just solder them right at the electrolytic leads.
Why buy 1$+ wima .1uF's ?? All the large resistors will be through- hole.
OS
Attachments
