http://www.passlabs.com
Check in the download section under service manuals.
There you can find the schematic and how it works.
Check in the download section under service manuals.
There you can find the schematic and how it works.
Aleph output stage is not power follower
The Aleph output stage is not a power follower. In fact, the output stage of an Aleph is essentially a Zen with active load. The output has voltage gain.
It would be possible to set up the output stage as a follower -- this would require inversion of transistor polarity and much higher voltage gain in input stage, probably a different input stage to accommodate that drive voltage is referenced to 0V instead of negative rail. Phase would also be inverted relative to the current configuration.
The X series, the old Stasis series and most other transistor designs use followers in output stage -- be they BJT, IBGT or MOSFET based. I strongly suspect Levinson and Krell do the same.
Petter
The Aleph output stage is not a power follower. In fact, the output stage of an Aleph is essentially a Zen with active load. The output has voltage gain.
It would be possible to set up the output stage as a follower -- this would require inversion of transistor polarity and much higher voltage gain in input stage, probably a different input stage to accommodate that drive voltage is referenced to 0V instead of negative rail. Phase would also be inverted relative to the current configuration.
The X series, the old Stasis series and most other transistor designs use followers in output stage -- be they BJT, IBGT or MOSFET based. I strongly suspect Levinson and Krell do the same.
Petter
How to choose power FET's for audio
I don't know it all, but this is my experience from research:
According to Nelson Pass, you are better off with a capacitor which has a fairly constant or at least little-changing input capacitance. New devices are optimized for switching and not linear use. In fact, the guys over at IRF once told med that their products are not recommended for linear use ....
Pass has in the past used older devices (generation 3 in IRF nomenclature). It is not obvious that newer devices will perform any better. I have spent countless hours checking out datasheets in search for the "ideal" device. After having searched for a long time for:
1. Form factor preferably 247 or better, but 220's also evaluated.
2. Low thermal resistance from substrate to heatsink
3. Low capacitance (typically CRSS, depending on circuit)
4. Double check capacitance with capacitance graphs
5. Low Rds(on) -- but this is not really that important.
etc.
The longer I looked, the harder it became. In the end, I started looking for modern devices with relatively high Rds(on) since these seemed to be better in most ways. I am probably going to end up with 3rd genneration units.
With CRSS(on), it turns out that each manufacturer has a different way of measuring it, and it is very variable with applied voltages. Qg seems to be a better measurement to work from, but this is also a variable with applied voltage.
What I did find in modern units that appeared to be "best" although suffering in device form factor (being 220's) was:
HUF75309P3
HUF75637P3
But these devices suffered from poor thermals, and I was still not sure about their sonics (but they were certainly reasonably priced!).
IRF530 and variants of it don't seem all that bad either apart from thermals. What most people forget is that the thermal resistance increases significantly when you add the insulor + grease. This can be overcome using BeO insulators, but I have not found these for smaller than TO3 size transistors (Beware that BeO is toxic when sanded and can cause berylliosis similar to astbestosis if inhaled -- but it is safe when treated carefully). Last time I checked they were available from a Thermalloy distributor. Be aware that they are also illegal in many countries.
If you are interested in TO3 availability from Intersil, go to this site http://www.intersil.com/mosfets/to3.asp and if you are interested in same from IRF, some should still be available, certainly in the IRF or IRFP240 series and it's complementary 9240 -- http://irplus.irf.com/search/part-search/part-search.2.html?BUCKET=BEGIN&PARTIAL=%&PART_NAME=irfp240 -- these are "exactly" the same as 244.
Why do designers use older devices? Most high-end manufacturers rely on matching of components. Semiconductors have a huge manufacturing tolerance, and in order to obtain good prices as well as the ability to match without major loss, they end up purchasing in bulk. This moves the price point way down, and with a large stock they are easily able to utilize 90-100% of the devices that they have paid for. For an amateur it is not so easy -- first we have to pay full price, then we have to buy 2-3 times what we need to even begin getting reasonably matched sets.
So there you have it, and good luck selecting devices. In the end, you are probably going to be more affected by matching of devices and quality of power supply/construction than the device choice.
For the input stage, please do not insert a modern low-resistance unit as you don't need it, and you need the low capacitance. 9610 seems like a good choice -- buy more than you need and match them up.
In the end I will probably end up with IRFP240, as few as possible, and with BeO isolators which improve CRSS because they are thick.
Petter
[Edited by Petter on 01-29-2001 at 02:23 PM]
I don't know it all, but this is my experience from research:
According to Nelson Pass, you are better off with a capacitor which has a fairly constant or at least little-changing input capacitance. New devices are optimized for switching and not linear use. In fact, the guys over at IRF once told med that their products are not recommended for linear use ....
Pass has in the past used older devices (generation 3 in IRF nomenclature). It is not obvious that newer devices will perform any better. I have spent countless hours checking out datasheets in search for the "ideal" device. After having searched for a long time for:
1. Form factor preferably 247 or better, but 220's also evaluated.
2. Low thermal resistance from substrate to heatsink
3. Low capacitance (typically CRSS, depending on circuit)
4. Double check capacitance with capacitance graphs
5. Low Rds(on) -- but this is not really that important.
etc.
The longer I looked, the harder it became. In the end, I started looking for modern devices with relatively high Rds(on) since these seemed to be better in most ways. I am probably going to end up with 3rd genneration units.
With CRSS(on), it turns out that each manufacturer has a different way of measuring it, and it is very variable with applied voltages. Qg seems to be a better measurement to work from, but this is also a variable with applied voltage.
What I did find in modern units that appeared to be "best" although suffering in device form factor (being 220's) was:
HUF75309P3
HUF75637P3
But these devices suffered from poor thermals, and I was still not sure about their sonics (but they were certainly reasonably priced!).
IRF530 and variants of it don't seem all that bad either apart from thermals. What most people forget is that the thermal resistance increases significantly when you add the insulor + grease. This can be overcome using BeO insulators, but I have not found these for smaller than TO3 size transistors (Beware that BeO is toxic when sanded and can cause berylliosis similar to astbestosis if inhaled -- but it is safe when treated carefully). Last time I checked they were available from a Thermalloy distributor. Be aware that they are also illegal in many countries.
If you are interested in TO3 availability from Intersil, go to this site http://www.intersil.com/mosfets/to3.asp and if you are interested in same from IRF, some should still be available, certainly in the IRF or IRFP240 series and it's complementary 9240 -- http://irplus.irf.com/search/part-search/part-search.2.html?BUCKET=BEGIN&PARTIAL=%&PART_NAME=irfp240 -- these are "exactly" the same as 244.
Why do designers use older devices? Most high-end manufacturers rely on matching of components. Semiconductors have a huge manufacturing tolerance, and in order to obtain good prices as well as the ability to match without major loss, they end up purchasing in bulk. This moves the price point way down, and with a large stock they are easily able to utilize 90-100% of the devices that they have paid for. For an amateur it is not so easy -- first we have to pay full price, then we have to buy 2-3 times what we need to even begin getting reasonably matched sets.
So there you have it, and good luck selecting devices. In the end, you are probably going to be more affected by matching of devices and quality of power supply/construction than the device choice.
For the input stage, please do not insert a modern low-resistance unit as you don't need it, and you need the low capacitance. 9610 seems like a good choice -- buy more than you need and match them up.
In the end I will probably end up with IRFP240, as few as possible, and with BeO isolators which improve CRSS because they are thick.
Petter
[Edited by Petter on 01-29-2001 at 02:23 PM]
IRF9610
In the Volksamp circuit there are two resistors near the input mosfets, marked with (*). They are probably for adjusting the gain of the two devices. What do you think? Is this a good approach or it's better to match the devices gain as Nelson says in his DIY OP amp article.
[Edited by Asen on 01-29-2001 at 04:57 PM]
In the Volksamp circuit there are two resistors near the input mosfets, marked with (*). They are probably for adjusting the gain of the two devices. What do you think? Is this a good approach or it's better to match the devices gain as Nelson says in his DIY OP amp article.
[Edited by Asen on 01-29-2001 at 04:57 PM]
Hi,
Promitheus: The IRFP150 datasheet from IRF says the IRFP150 is only a mounting hole isolated device and not a fully isolated one.
Usually fully isolated devices are not capable of high power capacities as you are talking about (>100W) and a design based in fully isolated devices needs much more heat dissipation help or more devices than the not fully isolated based. fully isolated devices have high thermal junction to case resistance.
I think T0247AC devices will be a very good choice if the size of heat sinks and mounting issues are considered.
There are some japanese devices using different encapsulation that could be interesting for audio designs (from Toshiba and Sanken). They usually have less thermal resistance than TO247 devices. I don't know if mosfets types are available.
Regards
Promitheus: The IRFP150 datasheet from IRF says the IRFP150 is only a mounting hole isolated device and not a fully isolated one.
Usually fully isolated devices are not capable of high power capacities as you are talking about (>100W) and a design based in fully isolated devices needs much more heat dissipation help or more devices than the not fully isolated based. fully isolated devices have high thermal junction to case resistance.
I think T0247AC devices will be a very good choice if the size of heat sinks and mounting issues are considered.
There are some japanese devices using different encapsulation that could be interesting for audio designs (from Toshiba and Sanken). They usually have less thermal resistance than TO247 devices. I don't know if mosfets types are available.
Regards
Newer FETs and low Rds on
There have been several questions about using newer generation fets and lower Rds on parts.While IR has continued to come out with "better" versions of the old standby parts these are not suitable for audio.The main thrust for improvement in these fets are for switching applications.The improved parts have lower Rds on for a given die area.For a given drain current they can now use a smaller die.This means lower cost.Unfortunately for audio it means less power handling.The new and improved IRF540 is rated at 94 watts where the original was rated at 125 watts.
There have been several questions about using newer generation fets and lower Rds on parts.While IR has continued to come out with "better" versions of the old standby parts these are not suitable for audio.The main thrust for improvement in these fets are for switching applications.The improved parts have lower Rds on for a given die area.For a given drain current they can now use a smaller die.This means lower cost.Unfortunately for audio it means less power handling.The new and improved IRF540 is rated at 94 watts where the original was rated at 125 watts.
blmn : I didn´t see that about the isolated hole but you are probably right. The last FETs I used were the irf250 and they were not fully isolated. They were to247 and I had no problem with heat disipation.
djk : I haven´t used any of the new ones.
Which ones do you think are good for audio ?
About the Japanese ones I know the Sangen BJT´s. I don´t have a clue if there are Mosfets too. One of the best FETs for audio are the Hitachi ones, but they are very expensive and the pin configuration is usually diferent from the IR.
Any suggestions are accepted.
djk : I haven´t used any of the new ones.
Which ones do you think are good for audio ?
About the Japanese ones I know the Sangen BJT´s. I don´t have a clue if there are Mosfets too. One of the best FETs for audio are the Hitachi ones, but they are very expensive and the pin configuration is usually diferent from the IR.
Any suggestions are accepted.
Petter,
The link below is the location of Hitachi's Power Mosfets for amplifier applications:
http://www.hitachi.co.jp/Sicd/English/Products/transise.htm
Hitachi devices are called "Lateral Mosfets", a different geometry to make this device. I don't know about the new units, but the old 2sk135/2sj50 pair was made using this technology. It seems to be the difference that made these devices very appreciated for amplifier designs.
An interesting comparative work about FETs and BJTs is in the chapter 10 of "Audio Power Amplifier Design Handbook" - Douglas Self is the author.
Regards
The link below is the location of Hitachi's Power Mosfets for amplifier applications:
http://www.hitachi.co.jp/Sicd/English/Products/transise.htm
Hitachi devices are called "Lateral Mosfets", a different geometry to make this device. I don't know about the new units, but the old 2sk135/2sj50 pair was made using this technology. It seems to be the difference that made these devices very appreciated for amplifier designs.
An interesting comparative work about FETs and BJTs is in the chapter 10 of "Audio Power Amplifier Design Handbook" - Douglas Self is the author.
Regards
Yeah the 2SK135 / 2SJ50 were very good. They are in TO-3 case. I tried to find them for the last amp I made but they were so expensive. They are out of production. So I found the IRFP250 and IRFP9250 at 1/7th of the price. I used them instead but since the pin lay out was diferent I had to make a different amp since I didn´t want to make new PCBs.
Hitachi has new FETs for audio in TO-247.
2SJ162 and 2SK1058 They are 2-3 times more expensive then IR or Harris.
That´s all I know about Hitachi.
Hitachi has new FETs for audio in TO-247.
2SJ162 and 2SK1058 They are 2-3 times more expensive then IR or Harris.
That´s all I know about Hitachi.
Petter,
The link below is the location of Hitachi's Power Mosfets for amplifier applications:
http://www.hitachi.co.jp/Sicd/English/Products/transise.htm
Hitachi devices are called "Lateral Mosfets", a different geometry to make this device. I don't know about the new units, but the old 2sk135/2sj50 pair was made using this technology. It seems to be the difference that made these devices very appreciated for amplifier designs.
An interesting comparative work about FETs and BJTs is in the chapter 10 of "Audio Power Amplifier Design Handbook" - Douglas Self is the author.
Regards
The link below is the location of Hitachi's Power Mosfets for amplifier applications:
http://www.hitachi.co.jp/Sicd/English/Products/transise.htm
Hitachi devices are called "Lateral Mosfets", a different geometry to make this device. I don't know about the new units, but the old 2sk135/2sj50 pair was made using this technology. It seems to be the difference that made these devices very appreciated for amplifier designs.
An interesting comparative work about FETs and BJTs is in the chapter 10 of "Audio Power Amplifier Design Handbook" - Douglas Self is the author.
Regards
Magnetek FETs
These are lateral fets like the Hitachi. http://www.semelab.co.uk/magnatec_latmos_pricing.htm They are direct replacement for the parts used in the Hafler DH500.They can be paralled without matching and do not exhibit the thermal runaway problems that the IR,Harris,all other D-mos type have.In the metal TO3 case the 250 watt units in 100 lot run about USD$9.00 each.There are also true complements available.While this is about 1.5X more than the IR and Harris plastic stuff on a cost per watt basis the advantages outweigh the cost in my mind.Also note that 500 watt packages that look like the big square solid state relay packs are available.
These are lateral fets like the Hitachi. http://www.semelab.co.uk/magnatec_latmos_pricing.htm They are direct replacement for the parts used in the Hafler DH500.They can be paralled without matching and do not exhibit the thermal runaway problems that the IR,Harris,all other D-mos type have.In the metal TO3 case the 250 watt units in 100 lot run about USD$9.00 each.There are also true complements available.While this is about 1.5X more than the IR and Harris plastic stuff on a cost per watt basis the advantages outweigh the cost in my mind.Also note that 500 watt packages that look like the big square solid state relay packs are available.
The Rds of the IR/Harris parts has a negative slope until about 5 amps.The Hitachi/Magnetec lateral mosfets slope positive at 100mA.The Vgs of the IR/Harris parts also has a strong temperature slope.To really match the IR/Harris types we need to match the Vgs at the actual drain current and opeating temperature we will be using.By design the Hitachi/Magnetec types are self-balancing over a wide range.The only safe way to use the IR/Harris types in a non class A circuit is to use a bias servo.Linear Technology makes such a device but it is expensive and they still recomend one bias IC per complementary pair of outputs.
Hi I am back !
I was just looking out for some heatsinks.
I had this wierd calculation and I don´t know how to explain it.
I found a heatsink 30cm*30cm*4cm rated 0.29 K/W at a cost of 160 DM. Around 80 USD.
I found some small ones 10cm*10cm*5cm rated 0.8 K/W at a cost of 12 DM (about 6 USD)
If I use 3 of them I have 0.267 K/W at a cost of 18 USD.
How can that be possible? That´s about 1/3 rd the heatsink with better cooling. What is wrong with the calculation?
I calculated in both situations 3 mosfets dissipating 20 watts each.
I was just looking out for some heatsinks.
I had this wierd calculation and I don´t know how to explain it.
I found a heatsink 30cm*30cm*4cm rated 0.29 K/W at a cost of 160 DM. Around 80 USD.
I found some small ones 10cm*10cm*5cm rated 0.8 K/W at a cost of 12 DM (about 6 USD)
If I use 3 of them I have 0.267 K/W at a cost of 18 USD.
How can that be possible? That´s about 1/3 rd the heatsink with better cooling. What is wrong with the calculation?
I calculated in both situations 3 mosfets dissipating 20 watts each.
I don´t remember exactly the first I said about but I found another one.
SK 56 200mm long is 300mm*40mm*200mm black rated 0.3 K/W,it has 29 parallel fins 40mm long and has around 50 USD.
SK 88 100mm long is 100mm*50mm*100mm black rated 0.8 K/W, not parallel fins ( the profil looks likes the ones aleph 30 from Volksamp )and has 6 USD.
I think the manufacturer is Fischer but the codes apply to others too.
SK 56 200mm long is 300mm*40mm*200mm black rated 0.3 K/W,it has 29 parallel fins 40mm long and has around 50 USD.
SK 88 100mm long is 100mm*50mm*100mm black rated 0.8 K/W, not parallel fins ( the profil looks likes the ones aleph 30 from Volksamp )and has 6 USD.
I think the manufacturer is Fischer but the codes apply to others too.
Looks like Fischer part numbers
Pricing on alu extrusions most of the time baffles me. The reason for this can be as varied as volume, stock, which are standard etc. Many manufacturers sell the same heatsinks under their own name even though they are only made in one place (one company, often competing). Some extrusion are hard to make (complex big ones typically). You might also try some of the surplus houses or make your own sinks (alu plates with spacers much like a radiator. When you go searching the web you will find at least one article on this.
Another good place to check (and they have software for looking at heatsinks is http://www.aavid.com/)
You could probably get cheaper pricing by buying in lengths of 1-2M and cut.
Pricing on alu extrusions most of the time baffles me. The reason for this can be as varied as volume, stock, which are standard etc. Many manufacturers sell the same heatsinks under their own name even though they are only made in one place (one company, often competing). Some extrusion are hard to make (complex big ones typically). You might also try some of the surplus houses or make your own sinks (alu plates with spacers much like a radiator. When you go searching the web you will find at least one article on this.
Another good place to check (and they have software for looking at heatsinks is http://www.aavid.com/)
You could probably get cheaper pricing by buying in lengths of 1-2M and cut.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.