I have fair dinkum done less reading in a library 😀
Keypunch;
The loss for a transistor N-channel output stage will be less than this N-channel Mosfet output stage. This is because these Mosfets require between 4 - 8 volts drive (Vgs) depending on the output current. For the positive rail Mosfets, this voltage differential dissappears as the Mosfet source approaches rail (at clipping) and so the best Vds that can be achieved is about the same as Vgs. The Mosfets on the negative rail have the luxury of lots of available +ve voltage and so do not suffer from this. This is why, without an auxillary voltage, this type of amp will clip assymetrically. In my amp I have designed the third stage to achieve a degree of symmetry but it's not perfect and in practical terms makes little difference to the sonic quality at or near clipping. If you look at my other design you will see the auxillary voltage I'm talking about.
A transistor output stage does not suffer from this (as much). It is a current driven device and the output capability of the device is a function of the available base current. A transistor's Vce at saturation is almost zero and this makes for a more efficient design.
For lower voltage design a transistor output stage is better.
My view of the MJE340/350 SOAR was for the voltages you were going to use.
The fuse chosen is based on experience. It needs to blow quickly in the event of a problem but must stay intact for all levels of "normal" output. Given the dynamics of music it is difficult to select the right fuse. In my view an extended burst of sine wave at full output should melt the fuse.
Fairchild and Motorola/OnSemi are both long standing quality manufacturers. As with many things in the electronics industry a fair bit of re-packaging and re-branding goes on so you just never know, the actual die might be exactly the same. In any case I would be comfortable using any transistor from this type of manufacturer (provided it wasn't a fake).
Cheers
Quasi
Keypunch;
The loss for a transistor N-channel output stage will be less than this N-channel Mosfet output stage. This is because these Mosfets require between 4 - 8 volts drive (Vgs) depending on the output current. For the positive rail Mosfets, this voltage differential dissappears as the Mosfet source approaches rail (at clipping) and so the best Vds that can be achieved is about the same as Vgs. The Mosfets on the negative rail have the luxury of lots of available +ve voltage and so do not suffer from this. This is why, without an auxillary voltage, this type of amp will clip assymetrically. In my amp I have designed the third stage to achieve a degree of symmetry but it's not perfect and in practical terms makes little difference to the sonic quality at or near clipping. If you look at my other design you will see the auxillary voltage I'm talking about.
A transistor output stage does not suffer from this (as much). It is a current driven device and the output capability of the device is a function of the available base current. A transistor's Vce at saturation is almost zero and this makes for a more efficient design.
For lower voltage design a transistor output stage is better.
My view of the MJE340/350 SOAR was for the voltages you were going to use.
The fuse chosen is based on experience. It needs to blow quickly in the event of a problem but must stay intact for all levels of "normal" output. Given the dynamics of music it is difficult to select the right fuse. In my view an extended burst of sine wave at full output should melt the fuse.
Fairchild and Motorola/OnSemi are both long standing quality manufacturers. As with many things in the electronics industry a fair bit of re-packaging and re-branding goes on so you just never know, the actual die might be exactly the same. In any case I would be comfortable using any transistor from this type of manufacturer (provided it wasn't a fake).
Cheers
Quasi
Hi quasi,
Thanks for your reply.
My prior reading over the months suggested more in the 4-5V Vgs. Your 4-8V makes more sense. The reading I had done seemed very universial to assume a 10V loss. I know there are other loss factors besides Vgs, but I do know what they are exactly, nor is it that important to me. As long as I have a basic idea of the loss range I do my calculations using a conservative side of the loss. I assume with a "perfect" PSU then the MOSFET losses would be close to 10V range.
I am aware of your effort to reduce the asymetrical clipping of this design. I know it is still asymetrical and feel given its close association to clipping that is of not a concern to me. My overbuilding should mean I really never approach the asymetrical region of operation of the amp. I am also aware of your other N-Channel design that uses the auxillary PSU and you do no consider the other design HiFi quality. I have read most of the trhead of your design with an auxillary PSU a few months back. Frankly speaking I do not think even with the early design of this amp using an auxillary supply was worth the cost and complexity for my needs. Maybe some time in future I might try an auxillary PSU as a curiousty to see what differences may arise. I suspect I likely will not do so.
Am I to assume the "transistor" (aka BiPolar) version output device losses are in order of 1-2V from what you expressed about the current oriented design "transistor" designs have?
Regarding the SOAR comment you made for the MJE340/350 was when you looked at the higher voltage/current handing ability.
I asked about fuse value sizing as I have observed a basic pattern to the dimensioned value vs aamplifier power so I deduced from the pattern a close aproximation to 8 ohms loads. I am aware of the rational of why, I was just trying to dimension to my rail voltages. I will be able to do a conservative calculation as start point. I suspect this will more than meet my modest listening/dynamic range needs. If not, well then I will up it in small steps to fine tune if the organ music turns out more demaning.
I asked about the KSE340/350 in case your being about this hobby and you rvast "quasi store of knowledge" you have some first hand, other builders, or via well regarded experienced others to give me a sense. I seem to read good things about Fairchild, but not knowing who created the MJE340/350 (I assume Motorola) and some devices have closely guarded designs I decided to ask and not assume. I recall simple specs such as as fT are not actually specified for some of the most common and popular devices was one indication of the more closely guarded designs. Reverse engineering can only tell engineers doing the reverse engineering only so much. I will not disclose everything in the manufacture of classic tubes. In the tubes case it was likely more a case of not knowing certain tubes would be so endearing and what the materials, gases, design and manufacture elements impact was to the final result. I know first had from disassembly of binary code for computers one cannot tell what reasons and intents were for various sections of the code. I have seen engineering bugs/flaws duplicated exactly to every cross "t" and dotted "i" from OEM designs and code/microcode reversed engineered due to inability to detect or understand in the reverse engineered item. So I asked the MJE vs KSE differences in case in your much longer audio/electronics travels and greater depth of knowledge you might have some insights to offer.
Rest asured the KSE340/350 I have are not fakes. Ergo my standoffish feeing to building a BiPolar output device based design due to fake issues. I have been shamed by this once alreadly despite me going in heads up. I discovered an additional method to detect fakes once I suspect I had one type being a fake. I just need to find a meter that performs tests more correctly. A test which deduces die size without having to break the casing open to see the die.
Regads,
John L. Males
Willowdale, Ontario
Canada
30 July 2006 (00:00 -) 01:12
Thanks for your reply.
My prior reading over the months suggested more in the 4-5V Vgs. Your 4-8V makes more sense. The reading I had done seemed very universial to assume a 10V loss. I know there are other loss factors besides Vgs, but I do know what they are exactly, nor is it that important to me. As long as I have a basic idea of the loss range I do my calculations using a conservative side of the loss. I assume with a "perfect" PSU then the MOSFET losses would be close to 10V range.
I am aware of your effort to reduce the asymetrical clipping of this design. I know it is still asymetrical and feel given its close association to clipping that is of not a concern to me. My overbuilding should mean I really never approach the asymetrical region of operation of the amp. I am also aware of your other N-Channel design that uses the auxillary PSU and you do no consider the other design HiFi quality. I have read most of the trhead of your design with an auxillary PSU a few months back. Frankly speaking I do not think even with the early design of this amp using an auxillary supply was worth the cost and complexity for my needs. Maybe some time in future I might try an auxillary PSU as a curiousty to see what differences may arise. I suspect I likely will not do so.
Am I to assume the "transistor" (aka BiPolar) version output device losses are in order of 1-2V from what you expressed about the current oriented design "transistor" designs have?
Regarding the SOAR comment you made for the MJE340/350 was when you looked at the higher voltage/current handing ability.
I asked about fuse value sizing as I have observed a basic pattern to the dimensioned value vs aamplifier power so I deduced from the pattern a close aproximation to 8 ohms loads. I am aware of the rational of why, I was just trying to dimension to my rail voltages. I will be able to do a conservative calculation as start point. I suspect this will more than meet my modest listening/dynamic range needs. If not, well then I will up it in small steps to fine tune if the organ music turns out more demaning.
I asked about the KSE340/350 in case your being about this hobby and you rvast "quasi store of knowledge" you have some first hand, other builders, or via well regarded experienced others to give me a sense. I seem to read good things about Fairchild, but not knowing who created the MJE340/350 (I assume Motorola) and some devices have closely guarded designs I decided to ask and not assume. I recall simple specs such as as fT are not actually specified for some of the most common and popular devices was one indication of the more closely guarded designs. Reverse engineering can only tell engineers doing the reverse engineering only so much. I will not disclose everything in the manufacture of classic tubes. In the tubes case it was likely more a case of not knowing certain tubes would be so endearing and what the materials, gases, design and manufacture elements impact was to the final result. I know first had from disassembly of binary code for computers one cannot tell what reasons and intents were for various sections of the code. I have seen engineering bugs/flaws duplicated exactly to every cross "t" and dotted "i" from OEM designs and code/microcode reversed engineered due to inability to detect or understand in the reverse engineered item. So I asked the MJE vs KSE differences in case in your much longer audio/electronics travels and greater depth of knowledge you might have some insights to offer.
Rest asured the KSE340/350 I have are not fakes. Ergo my standoffish feeing to building a BiPolar output device based design due to fake issues. I have been shamed by this once alreadly despite me going in heads up. I discovered an additional method to detect fakes once I suspect I had one type being a fake. I just need to find a meter that performs tests more correctly. A test which deduces die size without having to break the casing open to see the die.
Regads,
John L. Males
Willowdale, Ontario
Canada
30 July 2006 (00:00 -) 01:12
Keypunch
I think Mr. Males is writing a thesis on Quasi 😉 You may find when composing your posts it is sometimes helpful using a text editor to aid the reduction of typos etc. I often cut and paste from Word. But if you don't mind, I don't mind, just a suggestion.
I have more quasi coming out of the oven...soon with some nice pics 😀 This time I have an exaggerated power supply to appease Andrew and his Soar 🙂
I have built up a sense of comfort running Quasi without the DC protection circuit. I etched more boards without it and was wondering how folks felt about that. I see that Duong is not using the DC protection. Many purists swear against such circuitry in the path of the "precious audio signal" and I originally planned to use it but on a separate PCB to help prevent the case from reaching too deep as the main amplifier board is shorter without the circuit. The 10 Mosfet board is obviously quite long for an amplifier.
My quasi has been running 24/7 for weeks! I'm taking it to a big party soon. It will be nice to "warm" it up for once.
I'm beat...ZZZZZZZZZZZZZZ
Shawn.
I think Mr. Males is writing a thesis on Quasi 😉 You may find when composing your posts it is sometimes helpful using a text editor to aid the reduction of typos etc. I often cut and paste from Word. But if you don't mind, I don't mind, just a suggestion.
I have more quasi coming out of the oven...soon with some nice pics 😀 This time I have an exaggerated power supply to appease Andrew and his Soar 🙂
I have built up a sense of comfort running Quasi without the DC protection circuit. I etched more boards without it and was wondering how folks felt about that. I see that Duong is not using the DC protection. Many purists swear against such circuitry in the path of the "precious audio signal" and I originally planned to use it but on a separate PCB to help prevent the case from reaching too deep as the main amplifier board is shorter without the circuit. The 10 Mosfet board is obviously quite long for an amplifier.
My quasi has been running 24/7 for weeks! I'm taking it to a big party soon. It will be nice to "warm" it up for once.
I'm beat...ZZZZZZZZZZZZZZ
Shawn.
Hi Shawn,
Your suggestion is great, except I still make the same typos no matter what approach I use. I would not use Word if I was using Windows to do as you suggested, but a text editor better than Notepad. There are many free text editors available for Windows. All far better than Notepad. I use Linux 🙂 🙂 🙂 as it is more stable than Windows. Just the same, your suggestion is great and I have actually taken that approach only to find it does not reduce my typos 🙁(
My typos have taken turn for worse last few days as my fav keyboard died. The one I use now is just so different in where even the alpha keys are in different places and no famous "IBM click" to know if and when I struck a key!! The space bar on this keyboard is too short, 1/2 length what was on all prior keyboards for past 20 odd years. It will take a bit of time, but I will find me a real IBM keyboard and be able to return to keys in "normal" places.
As FYI, I think the only persons able to write a thesis on quasi are quasi, his parents, wife, any brothers or sisters quasi may have, maybe his in-laws (hiding form the baseball bat!), his kids, or even the cat! I definitely do not qualify to write the thesis on quasi. quasi will be more than happy to hear that. lol lol lol lol
Interesting, I would of thought your PSU was not the limiting element, but the number of devices you had to support the SOAR.
Shawn, I think it would be worth the extra ounce of protection to have the DC protection circuit. Initially quasi posted the module board without the DC protection. Many requested the DC protection so quasi added in the DC protection to the module board. I believe quasi's reasoning to initially not include the DC protection was to give builders a choice in what circuit they wanted to use fo priotection such as DC, short circuit, etc. These two layouts were well before the many requests for higher power versions, i.e. more output drivers.
For sure your comment about the board being long enough for 5 pairs of output devices before adding in the DC protection at end of board. I would think if you are interested in taking the one you built to a party to warm it up it may be wise to add in the DC protection before you do. Of course the final decision to add in the DC protection is yours.
I am not sure if you have heard of an Apt Holman preamplifier, but it has a relay to enable muting of the preamp outputs. I have not had a problem with the relay as yet and Tom Holman was very selective about a good signal path. If you select a good relay with excellent current handing ability I see no reason for the relay to dirty up the output signal. Besides, when used at parties who will notice any possible output signal problems due to a DC circuit protection relay? Then maybe all the people in the party will be arranged in a neat school like arrangement quietly and listen to your amp for the duration of the party? lol lol lol lol Just a thought, Shawn 😉
You play your music 24/7 at home? Even when you sleep? Even from such a large amplifier? I can just see you having nightmares hearing the music travel through the DC protection relay 😉 Just kidding. I could not resist.
Great to hear you have more quasi's "in the oven".
Happy dreams as your quasi amp keeps playing.
John L. Males
Willowdale, Ontario
Canada
30 July 2006 (02:48 -) 03:30
Your suggestion is great, except I still make the same typos no matter what approach I use. I would not use Word if I was using Windows to do as you suggested, but a text editor better than Notepad. There are many free text editors available for Windows. All far better than Notepad. I use Linux 🙂 🙂 🙂 as it is more stable than Windows. Just the same, your suggestion is great and I have actually taken that approach only to find it does not reduce my typos 🙁(
My typos have taken turn for worse last few days as my fav keyboard died. The one I use now is just so different in where even the alpha keys are in different places and no famous "IBM click" to know if and when I struck a key!! The space bar on this keyboard is too short, 1/2 length what was on all prior keyboards for past 20 odd years. It will take a bit of time, but I will find me a real IBM keyboard and be able to return to keys in "normal" places.
As FYI, I think the only persons able to write a thesis on quasi are quasi, his parents, wife, any brothers or sisters quasi may have, maybe his in-laws (hiding form the baseball bat!), his kids, or even the cat! I definitely do not qualify to write the thesis on quasi. quasi will be more than happy to hear that. lol lol lol lol
Interesting, I would of thought your PSU was not the limiting element, but the number of devices you had to support the SOAR.
Shawn, I think it would be worth the extra ounce of protection to have the DC protection circuit. Initially quasi posted the module board without the DC protection. Many requested the DC protection so quasi added in the DC protection to the module board. I believe quasi's reasoning to initially not include the DC protection was to give builders a choice in what circuit they wanted to use fo priotection such as DC, short circuit, etc. These two layouts were well before the many requests for higher power versions, i.e. more output drivers.
For sure your comment about the board being long enough for 5 pairs of output devices before adding in the DC protection at end of board. I would think if you are interested in taking the one you built to a party to warm it up it may be wise to add in the DC protection before you do. Of course the final decision to add in the DC protection is yours.
I am not sure if you have heard of an Apt Holman preamplifier, but it has a relay to enable muting of the preamp outputs. I have not had a problem with the relay as yet and Tom Holman was very selective about a good signal path. If you select a good relay with excellent current handing ability I see no reason for the relay to dirty up the output signal. Besides, when used at parties who will notice any possible output signal problems due to a DC circuit protection relay? Then maybe all the people in the party will be arranged in a neat school like arrangement quietly and listen to your amp for the duration of the party? lol lol lol lol Just a thought, Shawn 😉
You play your music 24/7 at home? Even when you sleep? Even from such a large amplifier? I can just see you having nightmares hearing the music travel through the DC protection relay 😉 Just kidding. I could not resist.
Great to hear you have more quasi's "in the oven".
Happy dreams as your quasi amp keeps playing.
John L. Males
Willowdale, Ontario
Canada
30 July 2006 (02:48 -) 03:30
DC protection;
It is easy to fall into a sense of security and avoid implementing DC protection. The world according to Quasi says Never Never...no wait a minute...Always Always use DC protection.
In less than a few seconds you will (guaranteed) lose all connected woofers, even really expensive ones.
Calculation 1 (8 ohm speaker); 70v rail into 6 ohms = 816 watts
Calculation 2 (4 ohm speaker); 70v rail into 3 ohms = 1633 watts
As the coil enamel starts to melt it gets worse and the current goes up and hopefully the fuses blow.
Skip on the short cct protection before you skip on DC protection. Fets are cheaper than speakers.
I have a right mind to fly over and take your amplifier away (and drink your beer).
Quasi
It is easy to fall into a sense of security and avoid implementing DC protection. The world according to Quasi says Never Never...no wait a minute...Always Always use DC protection.
In less than a few seconds you will (guaranteed) lose all connected woofers, even really expensive ones.
Calculation 1 (8 ohm speaker); 70v rail into 6 ohms = 816 watts
Calculation 2 (4 ohm speaker); 70v rail into 3 ohms = 1633 watts
As the coil enamel starts to melt it gets worse and the current goes up and hopefully the fuses blow.
Skip on the short cct protection before you skip on DC protection. Fets are cheaper than speakers.
I have a right mind to fly over and take your amplifier away (and drink your beer).
Quasi
Regarding DC protection and the signal path;
The DC protection circuit posted is designed to disconnect the speakers only and will not turn the amp off or take any other action. There is only a relay contact in the audio path and this is at output level. A few milliohms of contact area will make diddly difference here....audiophile tissues available by return post.
Quasi
The DC protection circuit posted is designed to disconnect the speakers only and will not turn the amp off or take any other action. There is only a relay contact in the audio path and this is at output level. A few milliohms of contact area will make diddly difference here....audiophile tissues available by return post.
Quasi
Quasi,
Just thought, is there any value to using a relay with more poles in the DC protection circuit to disconnect the module input as well?
Regards,
John L. Males
Willowdale, Ontario
Canada
30 July 2006 03:48
P.S. One of my record setting shortest posts 🙂 🙂 🙂
30 July 2006 03:52 Typo correction. jlm
Just thought, is there any value to using a relay with more poles in the DC protection circuit to disconnect the module input as well?
Regards,
John L. Males
Willowdale, Ontario
Canada
30 July 2006 03:48
P.S. One of my record setting shortest posts 🙂 🙂 🙂
30 July 2006 03:52 Typo correction. jlm
A two line post has gotta be a record for you John.
No more damage will be done to the amp module if audio input is still connected. The amp module is already badly broken. It is unlikely that the input would also lock to rail so any pre-amp connected will be quite safe. The input capacitor also provides protection up stream.
Cheers
Quasi
No more damage will be done to the amp module if audio input is still connected. The amp module is already badly broken. It is unlikely that the input would also lock to rail so any pre-amp connected will be quite safe. The input capacitor also provides protection up stream.
Cheers
Quasi
Hi Key,
Find the highest peak current into a resistive load equal to your normal speaker load.
Let's assume you have 4ohm speakers and 60Vpk output voltage (equivalent to 450W into 4r) Ipk=60/4=15Apk (ohms law and ignore reactive loadings)
Use a rail fuse of about half this peak current.
for 450W into 4r use F7A or F8A
From what you were saying in the previous posts you are not using Bensen's spreadsheet. It gives you the graph of temp corrected SOAR and shows the amplfier demands at various loadings. Be a devil, input the 15030 and 340 data and SEE the difference, if any.
the rule I use for rail fuses goes as follows
Find the highest peak current into a resistive load equal to your normal speaker load.
Let's assume you have 4ohm speakers and 60Vpk output voltage (equivalent to 450W into 4r) Ipk=60/4=15Apk (ohms law and ignore reactive loadings)
Use a rail fuse of about half this peak current.
for 450W into 4r use F7A or F8A
From what you were saying in the previous posts you are not using Bensen's spreadsheet. It gives you the graph of temp corrected SOAR and shows the amplfier demands at various loadings. Be a devil, input the 15030 and 340 data and SEE the difference, if any.
Hi,
the DC protection relay will need to be a beast (highly rated = english version) to cut the DC fault current when a faulty device shorts a supply rail to the speaker.
Most relays are rated at AC currents and when DC is quoted it is many times lower.
eg. 240Vac 10A, 12Vdc 2A.
70Vdc 12A is in a different ballpark (expense and size).
Using a multi pole will help if you parallel the poles.
I think it was Borbely that suggested reversing the pole connections to reduce erosion or some other problem compared to normal paralleling.
But do go and consult the manufaturers data sheet for the proposed relay and it's suitability for breaking high DC currents.
the DC protection relay will need to be a beast (highly rated = english version) to cut the DC fault current when a faulty device shorts a supply rail to the speaker.
Most relays are rated at AC currents and when DC is quoted it is many times lower.
eg. 240Vac 10A, 12Vdc 2A.
70Vdc 12A is in a different ballpark (expense and size).
Using a multi pole will help if you parallel the poles.
I think it was Borbely that suggested reversing the pole connections to reduce erosion or some other problem compared to normal paralleling.
But do go and consult the manufaturers data sheet for the proposed relay and it's suitability for breaking high DC currents.
Hi,
Key's two liner was actually a one liner.
The word wrap facility made it look like a two liner on low resolution monitors.
Definitely a record.😉
Key's two liner was actually a one liner.
The word wrap facility made it look like a two liner on low resolution monitors.
Definitely a record.😉
2 channel DC detect PCB
Greetings friends,
Here's a 2 channel DC detect PCB based on the schematic posted in this thread. This has not been built and tested. Please check layout before construction.
Cheers
Quasi
Greetings friends,
Here's a 2 channel DC detect PCB based on the schematic posted in this thread. This has not been built and tested. Please check layout before construction.
Cheers
Quasi
AndrewT said:
By jingo you're right AndrewT. When I close my favourites list John's two-liner becomes a one-liner. A confirmed record indeed.
Cheers
MJE340/350
Hi,
If you are after faster devices than MJE340/350 try KSC1220A / KSA2690A. These are Fairchild equivalents of well known Toshiba devices 2sa1930/2sc5171. Toshiba devices are difficult to get but Mouser has Fairchild devices at reasonable price so I use these. They can also be used as a replacement of Hitachi 2sb649a/2sd669a drivers (not in production any more) favoured by Randy Sloan.
cheers,
Hi,
If you are after faster devices than MJE340/350 try KSC1220A / KSA2690A. These are Fairchild equivalents of well known Toshiba devices 2sa1930/2sc5171. Toshiba devices are difficult to get but Mouser has Fairchild devices at reasonable price so I use these. They can also be used as a replacement of Hitachi 2sb649a/2sd669a drivers (not in production any more) favoured by Randy Sloan.
cheers,
Hi,
take care with the 649/669, 1220/2690 and 1930/5171 recommendation.
They have super specs EXCEPT @ higher voltages.
The latter two have a knee in the SOA curve at 20V or less.
By the time you ask them to pass current @ 60Vce to 80Vce they are crippled in comparison 340/350.
If you are working with +-50Vrails then any of these will work well. You can stretch 649/669 to about+-60Vrail.
take care with the 649/669, 1220/2690 and 1930/5171 recommendation.
They have super specs EXCEPT @ higher voltages.
The latter two have a knee in the SOA curve at 20V or less.
By the time you ask them to pass current @ 60Vce to 80Vce they are crippled in comparison 340/350.
If you are working with +-50Vrails then any of these will work well. You can stretch 649/669 to about+-60Vrail.
mje340/350
Hi,
Good point Andrew. It all depends on operational collector currents required and of course on rail voltages. For collector currents up to about 250mA (maybe even 300mA) I'd use KSA/KSC devices for rail voltages coming close to their CE voltage limits - say up to about +/-70V. For higher collector currents and high rail voltages MJE340/350 or other devices should be considered.
The main advantage of the KSC/2sb/2sa family of transistors is their much better linearity (Hfe against collector current) than that of MJE, low Cob capacitance and speed - of course all within their safe operating area.
I use these devices in headphone amps (no problem with high rail voltages) and Power Amps with rail voltages up to +/- 56V. As for the last 25 years I've been building only active systems (with one exception only) so I do not need very high PS voltages and forgot that others go higher in this respect. For sound quality nothing beats active system (all other things equal) and this is the first upgrade one should consider when pursuing sonic fidelity. So instead of a pair of quasi's amps with high voltage PS consider building four with roughly the same number of output devices but at lower PS plus a good active crossover. Improvement guaranteed.
cheers.
Hi,
Good point Andrew. It all depends on operational collector currents required and of course on rail voltages. For collector currents up to about 250mA (maybe even 300mA) I'd use KSA/KSC devices for rail voltages coming close to their CE voltage limits - say up to about +/-70V. For higher collector currents and high rail voltages MJE340/350 or other devices should be considered.
The main advantage of the KSC/2sb/2sa family of transistors is their much better linearity (Hfe against collector current) than that of MJE, low Cob capacitance and speed - of course all within their safe operating area.
I use these devices in headphone amps (no problem with high rail voltages) and Power Amps with rail voltages up to +/- 56V. As for the last 25 years I've been building only active systems (with one exception only) so I do not need very high PS voltages and forgot that others go higher in this respect. For sound quality nothing beats active system (all other things equal) and this is the first upgrade one should consider when pursuing sonic fidelity. So instead of a pair of quasi's amps with high voltage PS consider building four with roughly the same number of output devices but at lower PS plus a good active crossover. Improvement guaranteed.
cheers.
MJE340/350
Hi,
Small correction to my previous mail - it should be "PULSE collector currents up to 250mA...". These expected pulses should not exceed safe operating limits of KSA/KSC transistors in this case.
In practice, as long as PS voltages are not very high and amps driven hard with large signals, collector currents of predrivers are not very high. But true, it is important to analyse carefully specs of transistors against design objectives before choosing them.
cheers,
Hi,
Small correction to my previous mail - it should be "PULSE collector currents up to 250mA...". These expected pulses should not exceed safe operating limits of KSA/KSC transistors in this case.
In practice, as long as PS voltages are not very high and amps driven hard with large signals, collector currents of predrivers are not very high. But true, it is important to analyse carefully specs of transistors against design objectives before choosing them.
cheers,
Hi,
A1220 - 150mADC - 250mA 1mS,
B649 - 143mA - ?,
A1930 - 50mA - 270mA,
E340 - 260mA - 900mA,
E15030 - 140mA - ~1500 to 2000mA.
Then apply a temperature derating (Tc>25degC) and you have SOA ratings.
I do not recommend using any of the alternatives at 70Vce, even using the non repetitive 1mS rating. One would be seriously risking overstressing these drivers.
Using any of these drivers @ +-50V is OK, +-60V is pushing reliability, +-65V is on the point of repeated failures with real speakers (reactive load).
The 70Vce limits @25degC are
A1220 - 150mADC - 250mA 1mS,
B649 - 143mA - ?,
A1930 - 50mA - 270mA,
E340 - 260mA - 900mA,
E15030 - 140mA - ~1500 to 2000mA.
Then apply a temperature derating (Tc>25degC) and you have SOA ratings.
I do not recommend using any of the alternatives at 70Vce, even using the non repetitive 1mS rating. One would be seriously risking overstressing these drivers.
Using any of these drivers @ +-50V is OK, +-60V is pushing reliability, +-65V is on the point of repeated failures with real speakers (reactive load).