I have a few.rad 26260 only 150mm , I increased total surfece area with few 50 * 160 mm on botto& m , I tested the semelab 6331-R & 9411-R
the njw0281 njw 0302 as drivers bias is high lower pair 0.645 upper 0.647 . I got sanken drivers 2sa1186 and the 2837 metal case as semelabs .picked.a pair of little highe ddmm diode measurment real close to semelabs . Is this about right or anybody with other ideas and knowledge.
Njw version easy to build but the outputs high. Cel./W i think it was 0.78 against 0.32 how to compare heatsinks modushop 300*200 0.2 C & cheap Fisher sk56 good build quality worse dataspecs . 160mm * 100 mm 0.7C/W if I combinkme 3*3 about the s.ame cost as Sk56/200 but it would be 480mm * 300mm how much do you loose connenting several heatssinks. My first idea was getting 1M rad 26260 profile from as speaker stand amp put the trafos near floor. But jwilliem said it would not be a good fit . Too narrow , he said . OK 2M but thats 532 or split right 1M my sub s would look small
Lähetetty minun H60-L04 laitteesta Tapatalkilla
the njw0281 njw 0302 as drivers bias is high lower pair 0.645 upper 0.647 . I got sanken drivers 2sa1186 and the 2837 metal case as semelabs .picked.a pair of little highe ddmm diode measurment real close to semelabs . Is this about right or anybody with other ideas and knowledge.
Njw version easy to build but the outputs high. Cel./W i think it was 0.78 against 0.32 how to compare heatsinks modushop 300*200 0.2 C & cheap Fisher sk56 good build quality worse dataspecs . 160mm * 100 mm 0.7C/W if I combinkme 3*3 about the s.ame cost as Sk56/200 but it would be 480mm * 300mm how much do you loose connenting several heatssinks. My first idea was getting 1M rad 26260 profile from as speaker stand amp put the trafos near floor. But jwilliem said it would not be a good fit . Too narrow , he said . OK 2M but thats 532 or split right 1M my sub s would look small
Lähetetty minun H60-L04 laitteesta Tapatalkilla
Silver mica vs smd 1206 caps ?
Bought smd,s a few values that from 22pF to 680pF each 100 for my hobby.
10 pcs silvermica price or 1000-1500pcs smd why use mica , Is it a bad idea to jumper with twisted triplet to get output power device to npn/pnp/npn/pnp/npn & pnp/npn/pnp/npn/pnp 2 rows could this help thermal comp. with servo . What do you think of 2pcs H profile heatsinks 352mm that holds the pcb ips & ops a great find or not for 20€ .
Lähetetty minun H60-L04 laitteesta Tapatalkilla
Bought smd,s a few values that from 22pF to 680pF each 100 for my hobby.
10 pcs silvermica price or 1000-1500pcs smd why use mica , Is it a bad idea to jumper with twisted triplet to get output power device to npn/pnp/npn/pnp/npn & pnp/npn/pnp/npn/pnp 2 rows could this help thermal comp. with servo . What do you think of 2pcs H profile heatsinks 352mm that holds the pcb ips & ops a great find or not for 20€ .
Lähetetty minun H60-L04 laitteesta Tapatalkilla
I 100% agree.
I have experience, when I tested my amplifier with my friend speaker, I noticed the high slightly harsh only in several minute. But when I tested it with my speaker, I noticed it almost in 30 minutes. My friend speaker better than mine and much more expensive.
It would be rare for any "premium" component to make an amplifier perform better.
Most properly designed electronics will work perfectly well using specifcation compliant components.
i.e. don't use fake components, or reject components, or specially printed lables on low quality (non compliant)components.
Most properly designed electronics will work perfectly well using specifcation compliant components.
i.e. don't use fake components, or reject components, or specially printed lables on low quality (non compliant)components.
If I recall correctly the devices were the Magnatec (Semelab / TT Electronics) MG633x / MG941x devices as found here:
Complementary Bipolar Power Transistors for High Power Audio
These are fast large SOA devices that have been referred to as 'premium' devices. They are not cheap nor widely available. Would such a device would sound better? Doubt it. More output power from a given number of devices (due to the aforementioned larger than typical SOA) assuming the power supply et al can accomodate? Yes.
Complementary Bipolar Power Transistors for High Power Audio
These are fast large SOA devices that have been referred to as 'premium' devices. They are not cheap nor widely available. Would such a device would sound better? Doubt it. More output power from a given number of devices (due to the aforementioned larger than typical SOA) assuming the power supply et al can accomodate? Yes.
semelab
I haven't dared the mg 9413 & mg 6333 , always seems there was a mag9413w & mg6333 , hfe class a or b , I opted for the mg 9411-r mg6331-r (price was the factor) , but feel free and buy 25 each premium at 623,10€ .
I'm poor, sol stoped smoking cigarets 5 years ago but can't seem to afford good vape of magic smoke either.
Last buy was sanken 10 pcs 2sa1295 & 10pcs 2sc3264 each and drivers for another amp 10pcs 2sa1668 , 10pcs 2sc4382 for 93€
I haven't dared the mg 9413 & mg 6333 , always seems there was a mag9413w & mg6333 , hfe class a or b , I opted for the mg 9411-r mg6331-r (price was the factor) , but feel free and buy 25 each premium at 623,10€ .
I'm poor, sol stoped smoking cigarets 5 years ago but can't seem to afford good vape of magic smoke either.
Last buy was sanken 10 pcs 2sa1295 & 10pcs 2sc3264 each and drivers for another amp 10pcs 2sa1668 , 10pcs 2sc4382 for 93€
Transistor choices
Since I realize my old output transistor list doesn't take into consideration the new arc welder layout, I have updated it slightly. Even though MOSFETs are now an official choice for output devices. I will still skip over them for now. Bipolar transistors are by far the more popular choice for this design.
No matter what, get your transistors from an authorized distributor! No eBay, Alibaba, or Amazon. The odds of getting real devices are not in your favor.
Sanken MT-200 packages are extra wide (37mm). This large package allows much better heat dissipation than smaller packages every other manufacturer currently uses. This gives these transistors a huge advantage in SOA.
The new arc welder layout was redesigned for these in mind. The old layout didn't have enough room to fit all five pair of these at once.
The SOA graphs in the datasheets are a much more accurate way to determine power handling for our application than the wattage rating.
Always check both the NPN and PNP compliment's data sheet. Sometimes one of them will have a nasty surprise not present in it's compliments ratings.
The bigger Semelab devices have the best SOA compared to everyone else, however they aren't a very large manufacturer. Therefore they can be hard to come by and are relatively expensive.
Sanken is the next best with their massive MT-200 package transistors. Thankfully these are easier to get now than they were in the past. Sanken has finally caught up and most authorized distributors have stock.
On Semi MJL4281/MJL4302 is the next best choice for output transistors that are easy to get.
While these choices are expensive, most of them can handle twice the power per device than 2SC5200/2SA1943 or NJW0281/NJW0302 at 80V.
I have put in bold outputs that have been successfully used in at least one SlewMaster build.
Toshiba: (avalable from Mouser and Digikey)
A more complete list is avalable here.
2SC5200/2SA1943 230V 150W 30MHz TO-264
2SC5242/2SA1962 230V 130W 30MHz TO-3P
Fairchild: (avaliable from Mouser and Digikey)
FJL4315/FJL4215 Toshiba 2SC5200/2SA1943 clone
FJA4313/FJA4213 Toshiba 2SC5242/2SA1962 clone
On Semi: (avalible from Mouser, Digikey, and Newark/Farnell)
NJW0281/NJW0302 250V 150W 30MHz TO-247
NJW3281/NJW1302 250V 200W 30MHz TO-247
MJL3281/MJL1302 250V 200W 30MHz TO-264
MJL4281/MJL4302 350V 230W 35MHz TO-264
Semelab: (avalible from Profusion and Newark/Farnell)
A more complete list is avalable here.
MG6330/MG9410 230V 200W 35MHz TO-3P
MG6330-R/MG9410-R 260V 200W 35MHz TO-3P
MG6331/MG9411 230V 300W 35MHz TO-3P
MG6331-R/MG9411-R 260V 300W 35MHz TO-3P
MAG6332/MAG9412 230V 400W 35MHz TO-264
MG6333-R/MG9413-R 260V 400W 35MHz TO-264
Sanken: (available from Digikey, Profusion, and Newark/Farnell)
A more complete list is available here.
2SC6145A/2SA2223A 260V 160W 35MHz MT-100/TO-3
2SC3263/2SA1294 230V 130W 35MHz MT-100/TO-3P
2SC3264/2SA1295 230V 200W 35MHz MT-200
2SC3519/2SA1386 160V 130W 40MHz MT-100/TO-3P (70V rails max)
2SC2922/2SA1216 160V 200W 40MHz MT-200 (70V rails max)
2SC2921/2SA1215 160V 150W 50MHz MT-200 (70V rails max)
2SC3284/2SA1303 150V 125W 50MHz MT-100/TO-3P (60V rails max)
While one could just use the same transistors for the drivers as the outputs there is some minor benefit of choosing smaller devices with lower Cob (parasitic capacitance). This is where 2SC5200/2SA1943 or NJW0281/NJW0302 would be a great choice. 2SC3263/2SA1294 has higher Cob, but has a faster fT for those concerned about speed. If you are only running 2 or 3 pairs then even smaller transistors can be used. MJE15032/MJE15033 and FJP5200/FJP1943 come to mind.
KSC3503/KSA1381 is by far the best choice for the pre-drivers. The only downside is they can be very difficult to find in a matching gain grade. No builders seem to have had any issues with mixing gain grades however.
There are many potential substitutes, Toshiba 2SC4793/2SA1837 being the next best choice. However, none of them have the ultra-low Cob of KSC3503/KSA1381.
Since I realize my old output transistor list doesn't take into consideration the new arc welder layout, I have updated it slightly. Even though MOSFETs are now an official choice for output devices. I will still skip over them for now. Bipolar transistors are by far the more popular choice for this design.
No matter what, get your transistors from an authorized distributor! No eBay, Alibaba, or Amazon. The odds of getting real devices are not in your favor.
Sanken MT-200 packages are extra wide (37mm). This large package allows much better heat dissipation than smaller packages every other manufacturer currently uses. This gives these transistors a huge advantage in SOA.
The new arc welder layout was redesigned for these in mind. The old layout didn't have enough room to fit all five pair of these at once.
The SOA graphs in the datasheets are a much more accurate way to determine power handling for our application than the wattage rating.
Always check both the NPN and PNP compliment's data sheet. Sometimes one of them will have a nasty surprise not present in it's compliments ratings.
The bigger Semelab devices have the best SOA compared to everyone else, however they aren't a very large manufacturer. Therefore they can be hard to come by and are relatively expensive.
Sanken is the next best with their massive MT-200 package transistors. Thankfully these are easier to get now than they were in the past. Sanken has finally caught up and most authorized distributors have stock.
On Semi MJL4281/MJL4302 is the next best choice for output transistors that are easy to get.
While these choices are expensive, most of them can handle twice the power per device than 2SC5200/2SA1943 or NJW0281/NJW0302 at 80V.
I have put in bold outputs that have been successfully used in at least one SlewMaster build.
Toshiba: (avalable from Mouser and Digikey)
A more complete list is avalable here.
2SC5200/2SA1943 230V 150W 30MHz TO-264
2SC5242/2SA1962 230V 130W 30MHz TO-3P
Fairchild: (avaliable from Mouser and Digikey)
FJL4315/FJL4215 Toshiba 2SC5200/2SA1943 clone
FJA4313/FJA4213 Toshiba 2SC5242/2SA1962 clone
On Semi: (avalible from Mouser, Digikey, and Newark/Farnell)
NJW0281/NJW0302 250V 150W 30MHz TO-247
NJW3281/NJW1302 250V 200W 30MHz TO-247
MJL3281/MJL1302 250V 200W 30MHz TO-264
MJL4281/MJL4302 350V 230W 35MHz TO-264
Semelab: (avalible from Profusion and Newark/Farnell)
A more complete list is avalable here.
MG6330/MG9410 230V 200W 35MHz TO-3P
MG6330-R/MG9410-R 260V 200W 35MHz TO-3P
MG6331/MG9411 230V 300W 35MHz TO-3P
MG6331-R/MG9411-R 260V 300W 35MHz TO-3P
MAG6332/MAG9412 230V 400W 35MHz TO-264
MG6333-R/MG9413-R 260V 400W 35MHz TO-264
Sanken: (available from Digikey, Profusion, and Newark/Farnell)
A more complete list is available here.
2SC6145A/2SA2223A 260V 160W 35MHz MT-100/TO-3
2SC3263/2SA1294 230V 130W 35MHz MT-100/TO-3P
2SC3264/2SA1295 230V 200W 35MHz MT-200
2SC3519/2SA1386 160V 130W 40MHz MT-100/TO-3P (70V rails max)
2SC2922/2SA1216 160V 200W 40MHz MT-200 (70V rails max)
2SC2921/2SA1215 160V 150W 50MHz MT-200 (70V rails max)
2SC3284/2SA1303 150V 125W 50MHz MT-100/TO-3P (60V rails max)
While one could just use the same transistors for the drivers as the outputs there is some minor benefit of choosing smaller devices with lower Cob (parasitic capacitance). This is where 2SC5200/2SA1943 or NJW0281/NJW0302 would be a great choice. 2SC3263/2SA1294 has higher Cob, but has a faster fT for those concerned about speed. If you are only running 2 or 3 pairs then even smaller transistors can be used. MJE15032/MJE15033 and FJP5200/FJP1943 come to mind.
KSC3503/KSA1381 is by far the best choice for the pre-drivers. The only downside is they can be very difficult to find in a matching gain grade. No builders seem to have had any issues with mixing gain grades however.
There are many potential substitutes, Toshiba 2SC4793/2SA1837 being the next best choice. However, none of them have the ultra-low Cob of KSC3503/KSA1381.
Thanks for the avaliability list. Big help! Is there a person who matches the input devices Bc550C and BC560C or the KSC1815 and KSC1845 for the Snoopy IPS? Are the PHILLIPS original bjts preferred to any other brand? I looked up an article by Walt Jung when he was building an AMPZILLA published in AUDIO AMATEUR and describes the procedure for matching. Also a Marshall Leach article in AUDIO magazine where he discusses matching input devices. To properly match how many of a sample would one have to have: 100? Thanks, Ray
I would avoid Phillips unless you can be certain they aren't fake! Phillips (now named NXP) hasn't made any BC550/560 in quite awhile, genuine stock has long been sold out. While it can be argued Phillips made the best BC5XX transistors, there is little chance finding any that aren't fake anymore.
There is a benefit to matching, but isn't required for the Slewmaster designs.
I consider just matching by hFE on a cheap DMM good enough. Even this method can be a bit too fiddly, body heat from your fingers can skew the readings.
I would start out with 100 minimum if I wanted to do matching. This isn't as expensive as it sounds when one looks at the quantity price breaks from most reputable distributors.
For minimum dc offset at the output, Q1 through Q4 should preferably have matched current gains. The current gains can be measured with a curve tracer or with a multimeter which has this capability. Ideally, all four transistors should be matched. If this cannot be achieved, the second choice is for Q1 and Q3 to be matched and Q2 and Q4 to to be matched. The third choice is for Q1 and Q2 to be matched and Q3 and Q4 to be matched. The typical dc offset at the amplifier output is less than 50 mV. If you are concerned about how well the transistors are matched,
Take a look at http://leachlegacy.ece.gatech.edu/lowtim/part2.html
A newer idea would maybe be these matched npn,pnp smd devices but you would have to make flying leads from the present pcb or do a re-design to pcb but that's another story, going to test this in the future when time and inspiration allows .
Lähetetty minun H60-L04 laitteesta Tapatalkilla
Take a look at http://leachlegacy.ece.gatech.edu/lowtim/part2.html
A newer idea would maybe be these matched npn,pnp smd devices but you would have to make flying leads from the present pcb or do a re-design to pcb but that's another story, going to test this in the future when time and inspiration allows .
Lähetetty minun H60-L04 laitteesta Tapatalkilla
Are you talking about Super NAD CFA, shown in post #9467?
Then DC offset will be kept within a few mV by DC servo, even if all the mentioned transistors will be totally unmatched.
For the minimum distortion, it's enough to match the ones within the pairs - Q1, Q2 and Q3, Q4.
Symmetry topology like Spooky or Symmetri amplifier is the best for bass reproduction. Recently, I tried spooky amplifier with some modification (use easily to find transistors here and use different compensation and others modification). I used 90VDC PSU, it will use for party amp.
In simulation, I get 0.008537% THD at 344W/8Ohm, 20kHz and slew rate about 94V/uS. Prototype was working but need PCB layout modification for better thermal management. Sound fantastic in low frequency and the mid and high frequency did not sound any harsh at all. Bias current is very low, about 10mA
It sound like a hi-fi amplifier rather than audio pro amplifier
In simulation, I get 0.008537% THD at 344W/8Ohm, 20kHz and slew rate about 94V/uS. Prototype was working but need PCB layout modification for better thermal management. Sound fantastic in low frequency and the mid and high frequency did not sound any harsh at all. Bias current is very low, about 10mA
It sound like a hi-fi amplifier rather than audio pro amplifier
Last edited:
This was the spooky ips pcb layout by jknuteman. Yes, vzaichenko is right .There is the servo . If it helps anyone,I done the (Q1,Q2) & (Q3,Q4) as there is requests of philips or generally bc550c bc560c as input transistors . 20-23 years ago amplifier worked easy really easy , with no servo and the two pairs hfe matched with DMM hfe tester. was enough.
Nattawa did a his version of spooky he had a big influence , then thimos tested OS spooky page 127 & 130page , page 334 > 338 Still4given did his OS version . Now to rio
Lähetetty minun H60-L04 laitteesta Tapatalkilla
Nattawa did a his version of spooky he had a big influence , then thimos tested OS spooky page 127 & 130page , page 334 > 338 Still4given did his OS version . Now to rio
Lähetetty minun H60-L04 laitteesta Tapatalkilla
Ostripper's website posts FINAL = Spooky V1.2 which does not use the super pair circuit, and instead goes back to the standard Hawksford VAS. Ostripper stated that this is his favorite Spooky... favorite bass amp. Your posted V1.3 appears to use a few different resistor values.
http://startfetch.com/ostripper/docs/Spooky/Spookyamp-V1.2-final/Spooky-V1.2-schema.gif
=========================
The Spooky V2 you attached uses the super-pair in the VAS circuit. The super-pair adds a local feed forward loop within the global loop of the whole IPS. The super-pair has some value with lower loop gain IPS and in CFA, and even the Spooky front end. BUT.. The super-pair adds instability, especially if all of the VAS stage transistors are not matched. Ostripper recommended using matched transistors arrays for super-pairs.
Last edited: