TIP142/147 or TIP35/36 on +-55V is probably not going to be very reliable... And you'll have to test them at 110V and use only the ones that pass the test.
I have the following types if you are interested, I've found it a bit hard to find suitable transistors for power amps from the large distributors here in Sweden:
FJL4215/FJL4315 (Fairchild 2SC5200/2SA1943)
FJA4313/FJA4213 (The same but in the little smaller TO3-P package instead of TO-264)
NJL3281D/NJL1302D Thermaltraks (Just got these from Arrow, haven't had time to play with them yet though)
MJE340/MJE350
MJE15032/MJE15033 - these are nice as drivers 🙂
2n5551/2n5401
MPSA18
I have the following types if you are interested, I've found it a bit hard to find suitable transistors for power amps from the large distributors here in Sweden:
FJL4215/FJL4315 (Fairchild 2SC5200/2SA1943)
FJA4313/FJA4213 (The same but in the little smaller TO3-P package instead of TO-264)
NJL3281D/NJL1302D Thermaltraks (Just got these from Arrow, haven't had time to play with them yet though)
MJE340/MJE350
MJE15032/MJE15033 - these are nice as drivers 🙂
2n5551/2n5401
MPSA18
Power BJT's for LME49810 and heatsinks
Thanks for your offer megajocke, I may take your offer later on. The reason for choosing TIP35&36C at first was that I had them on hand. I expect to get the TIP142/147 tomorrow so I'll test them first to see if the difference. I have two boards running with TIP35/36 since three weeks, so far there has not been any failure. I am looking more for high quality sound than high power, so I'll probably not bother with drivers for the MOSFET/LME49830 version, and if the TIP142/147 combo works out as expected I'll probably not add a driver stage on the BJT version either. My original thought was to stick to the KISS principle and use as few components as possible, in my mind the more transistors you add the more you will taint or even distort the sound. But I must say that Pansons measurements tell me different...
Panson, I have to trust you when you say that the chip gets "just warm" with that small heatsink! I use heatsinks 3 times the size of yours (home brew 30x30x20 of 2mm thick aluminum bracket) and the chip gets more than just warm. After 2 hrs of easy listening the heatsink is 65-70 degree's C which is more than I would like it to be. As a rule of thumb, 10C above 55C reduces life time expectancy to a quarter. National recommends a heatsink of 10 degr. C/W. I found it in the US and ordered 4 of them. I just learned the UPS shipping is 80 USD.... Consequently I am right now redoing the layout to accomodate a variety of heatsinks. Same thing for heatsinks as for most anything if you live i Sweden - hard to get and if you do its damn expensive!
Thanks for your offer megajocke, I may take your offer later on. The reason for choosing TIP35&36C at first was that I had them on hand. I expect to get the TIP142/147 tomorrow so I'll test them first to see if the difference. I have two boards running with TIP35/36 since three weeks, so far there has not been any failure. I am looking more for high quality sound than high power, so I'll probably not bother with drivers for the MOSFET/LME49830 version, and if the TIP142/147 combo works out as expected I'll probably not add a driver stage on the BJT version either. My original thought was to stick to the KISS principle and use as few components as possible, in my mind the more transistors you add the more you will taint or even distort the sound. But I must say that Pansons measurements tell me different...
Panson, I have to trust you when you say that the chip gets "just warm" with that small heatsink! I use heatsinks 3 times the size of yours (home brew 30x30x20 of 2mm thick aluminum bracket) and the chip gets more than just warm. After 2 hrs of easy listening the heatsink is 65-70 degree's C which is more than I would like it to be. As a rule of thumb, 10C above 55C reduces life time expectancy to a quarter. National recommends a heatsink of 10 degr. C/W. I found it in the US and ordered 4 of them. I just learned the UPS shipping is 80 USD.... Consequently I am right now redoing the layout to accomodate a variety of heatsinks. Same thing for heatsinks as for most anything if you live i Sweden - hard to get and if you do its damn expensive!
Re: Power BJT's for LME49810 and heatsinks
May be my test duration (about 30 min) was too short. Or because the my circuit has a driver? Today i will run it longer with high output level and see how hot it can be. thank for your sharing.
Here in hong kong we can get many parts inexpensively from electronic market in Shenzhen, China. Quality ... on your own risk!
😀
Segran said:
May be my test duration (about 30 min) was too short. Or because the my circuit has a driver? Today i will run it longer with high output level and see how hot it can be. thank for your sharing.
Here in hong kong we can get many parts inexpensively from electronic market in Shenzhen, China. Quality ... on your own risk!
😀
Hi Sergan
I run a test with 1 kHz, 50 W/8 Ohm for an hour. The little heat sink for LME49810 was at 52 degree C.
BTW, second part of my report is here http://www.audio-domain.com/power_amplifier_projects.htm .
Comments are welcome.
I run a test with 1 kHz, 50 W/8 Ohm for an hour. The little heat sink for LME49810 was at 52 degree C.
BTW, second part of my report is here http://www.audio-domain.com/power_amplifier_projects.htm .
Comments are welcome.
The tip 142/147 will have enough current gain at least, but I wouldn't trust their SOA is good enough, and you are outside their voltage rating - but if your DC protection is working and you fuse appropriately the worst that can happen is that you blow the output transistors and possibly (but unlikely) the driver chip.
Just using TIP35/36 will have much too low current gain (especially with 2k2 base resistors). Output peak current might be as low as 2A-4A or so with those if you factor in beta droop.
The driver chip won't need a very big heatsink.
Just using TIP35/36 will have much too low current gain (especially with 2k2 base resistors). Output peak current might be as low as 2A-4A or so with those if you factor in beta droop.
The driver chip won't need a very big heatsink.
TIP142/147 - a REALLY BAD CHOICE!
Got the TIP 142/147's today, put them onto the PCB, adjusted the Bias to a "reasonable" 20mA..... Sounded alright at low 1-5W over 8ohm.....Looked nice on the scope too, but not different from my own choice of TIP35/36. Turned up the volume (I played"Curiosity killed the cat") gardually, the so called power meters in the Marantz chassie peaking at 20W - and POP! That was on set of transistors, C-E totally short circuited... OK, I don't gice up easy so in came then next pair. Same story, but I increased power even more softly. After 6 minutes at around 10W - POP again... Sigh, the third pair was mounted, this time a went directly for broke and set the volume at where the clipping indicator just went off- 30 seconds and POP! That's it for the TIP 142/147's!
The speaker protection circuit proved to be working.....
There's absolutely no scientific or mathematics behind this, it was a pure practical exercise. Don't think it was my choice of music that killed the transistors, only the cat....
Megajocke, you where dead on the money about the SOR of TIP 142/147. At +/- 55VDC they are simply a REALLY BAD CHOICE.
I remounted the TIP 35/36's again and have been playing the crap out of my speakers (and mye ears...), after 3 hrs at full power (clipping indicators going like christmas lights in S-pore) the amp is still playing!. Thats it for me, I will stick with my first choice until I have built the 49830/IRFP240/9240 set.
Megajocke, Do you have any recommendation for the 2k2 base resistors? I'm not sure if you mean I should increase or decrease..... I'm also interested in you offer for output transistors, specially the FJL's - if they will work without adding a driver stage. I don't feel like redoin the PCB's right now...
Panson, what driver transistors are you using? I haven't found any spec of those in your excellent documentation.
Got the TIP 142/147's today, put them onto the PCB, adjusted the Bias to a "reasonable" 20mA..... Sounded alright at low 1-5W over 8ohm.....Looked nice on the scope too, but not different from my own choice of TIP35/36. Turned up the volume (I played"Curiosity killed the cat") gardually, the so called power meters in the Marantz chassie peaking at 20W - and POP! That was on set of transistors, C-E totally short circuited... OK, I don't gice up easy so in came then next pair. Same story, but I increased power even more softly. After 6 minutes at around 10W - POP again... Sigh, the third pair was mounted, this time a went directly for broke and set the volume at where the clipping indicator just went off- 30 seconds and POP! That's it for the TIP 142/147's!
The speaker protection circuit proved to be working.....
There's absolutely no scientific or mathematics behind this, it was a pure practical exercise. Don't think it was my choice of music that killed the transistors, only the cat....
Megajocke, you where dead on the money about the SOR of TIP 142/147. At +/- 55VDC they are simply a REALLY BAD CHOICE.
I remounted the TIP 35/36's again and have been playing the crap out of my speakers (and mye ears...), after 3 hrs at full power (clipping indicators going like christmas lights in S-pore) the amp is still playing!. Thats it for me, I will stick with my first choice until I have built the 49830/IRFP240/9240 set.
Megajocke, Do you have any recommendation for the 2k2 base resistors? I'm not sure if you mean I should increase or decrease..... I'm also interested in you offer for output transistors, specially the FJL's - if they will work without adding a driver stage. I don't feel like redoin the PCB's right now...
Panson, what driver transistors are you using? I haven't found any spec of those in your excellent documentation.
fredlock said:
Hi fredlock,
Thanks for your interest! Sorry, the present PCB is out of stock. Meanwhile, I am working on new layout. Would you mind to wait for a while?
You will need to use a driver stage or darlington output devices
, but at +-55V there aren't that many suitable darlington outputs. I believe Sanken has some types with good SOA but I don't know their part numbers.
Even with a beta of 100 the 50mA drive will only allow 5A of output current, not even enough for driving 8 ohms to full power. As AndrewT already said optimally you should have output current capability of about 2-3 times vpeak/impedance - that is at least 15-20A. Isn't it possible to put drivers on the heatsink and connect them with short wires?
2k2 base resistors will drop 2.2V for every mA of base current. In the current setup with TIP35/TIP36 this means the amplifier will clip due to the voltage drop over these resistors already at 13W output! It is equivalent to having 22 ohm emitter resistors! 2k2 seems to be on the high side even for darlingtons.
The 1.2k resistor in series with the vbe multiplier will probably have to be removed to get bias stable. What's the thinking behind it? Also R14 seems very low in value letting almost all current go through the divider instead of through the transistor.
BD139/BD140 are not suitable as drivers at +-55V, they are only 80V Vce while 110V is needed. The MJE150xx, MJE34x or 2SC4793 and its complementary are more suitable.
You can probably use the PCB unmodified with IRFP mosfets though if you don't want to add drivers. 2.2k resistors are probably still too high, and the 100pf capacitors from G-D seem counterproductive. This is assuming the driver chip can have bias voltage high enough for the fets. (check datasheet) You should probably reduce the two 1k resistors that divides the voltage for feedback to get higher standing current too.
, but at +-55V there aren't that many suitable darlington outputs. I believe Sanken has some types with good SOA but I don't know their part numbers.
Even with a beta of 100 the 50mA drive will only allow 5A of output current, not even enough for driving 8 ohms to full power. As AndrewT already said optimally you should have output current capability of about 2-3 times vpeak/impedance - that is at least 15-20A. Isn't it possible to put drivers on the heatsink and connect them with short wires?
2k2 base resistors will drop 2.2V for every mA of base current. In the current setup with TIP35/TIP36 this means the amplifier will clip due to the voltage drop over these resistors already at 13W output! It is equivalent to having 22 ohm emitter resistors! 2k2 seems to be on the high side even for darlingtons.
The 1.2k resistor in series with the vbe multiplier will probably have to be removed to get bias stable. What's the thinking behind it? Also R14 seems very low in value letting almost all current go through the divider instead of through the transistor.
BD139/BD140 are not suitable as drivers at +-55V, they are only 80V Vce while 110V is needed. The MJE150xx, MJE34x or 2SC4793 and its complementary are more suitable.
You can probably use the PCB unmodified with IRFP mosfets though if you don't want to add drivers. 2.2k resistors are probably still too high, and the 100pf capacitors from G-D seem counterproductive. This is assuming the driver chip can have bias voltage high enough for the fets. (check datasheet) You should probably reduce the two 1k resistors that divides the voltage for feedback to get higher standing current too.
Thanks again Megajocke for your valuable input. The purpose of this first version was to get 2x70-80W amps to revive my burmed out Marantz SR-6000DC. I was ignorant to think that it may produce as much as 100W, but I have learned a lot from you all, using this thread. The mission with the Marantz is accomplished, it sound better than ever before inspite of the obvious shortcomings in delivering real power.
For the next version I have quite different obejctives however, but still satisfied by 100-130W/8ohm. Thats where the MOSFET and 49830 comes into the picture. I got all the parts ans only need to etch a couple of PCB's to assemble them. The layout is completed, this time I have put the speaker protection for both channels on a separate PCB. This time I don't have the same space limitations for the PCB's.
As a third step I will go back and redo the 49810/BJT PCB to include drivers. I have ordered 2SA1943 & 2SC5200 but haven't found better drivers than MJE340/350 or possibly MJE15032/15033. Maybe I can buy them from you?
For the 2k2 RB resistors, I started out on the prototype with 100R, but I couldn't get the bias down to anything reasonable. It was several hundred mA. I have tried with 1K as well, but still the bias is too high. With 2k2 I can adjust the bias from 10-50mA.
I realize that I'm limiting not only bias current but also output current, but this was the only way I could make it work at the time. But I can assure you that there is no clipping until 90Vpp on the outout, checked this with my scope. The power supply rails drops about 5 Volts each at this load, and an additional drop of 5 volt across each transistor I thnik is to be ecpected. Specially since I have no drivers... Maybe the chip actually can drive much more that what the spec's say?
The 1k2 and the 360R resistors as well as the 1+1k resistors in the voltage divider are simply cut and paste from National datasheet and Application notes. The same values are used in a number of publications. Panson uses 1k+1K in the voltage divider as well.
However, I just now realized that I have put the voltage divider on the wrong side of the RB's.... The voltage divider should sit pre RB's... Could that be the root cause to my problems with the Bias earlier? Seems unlikely. but I'll try it tomorrow.
In my MOSFET version I have put 120R as RB and 2x6k2 as voltage divder - as per National Application Note.
The 100pF G-D cap is put there as per another old MOSFET deisgn, I believe it comes from Elliot Sound Projectcs, but I canät find the reference right now. In what way do you see them as counterproductive?
For the next version I have quite different obejctives however, but still satisfied by 100-130W/8ohm. Thats where the MOSFET and 49830 comes into the picture. I got all the parts ans only need to etch a couple of PCB's to assemble them. The layout is completed, this time I have put the speaker protection for both channels on a separate PCB. This time I don't have the same space limitations for the PCB's.
As a third step I will go back and redo the 49810/BJT PCB to include drivers. I have ordered 2SA1943 & 2SC5200 but haven't found better drivers than MJE340/350 or possibly MJE15032/15033. Maybe I can buy them from you?
For the 2k2 RB resistors, I started out on the prototype with 100R, but I couldn't get the bias down to anything reasonable. It was several hundred mA. I have tried with 1K as well, but still the bias is too high. With 2k2 I can adjust the bias from 10-50mA.
I realize that I'm limiting not only bias current but also output current, but this was the only way I could make it work at the time. But I can assure you that there is no clipping until 90Vpp on the outout, checked this with my scope. The power supply rails drops about 5 Volts each at this load, and an additional drop of 5 volt across each transistor I thnik is to be ecpected. Specially since I have no drivers... Maybe the chip actually can drive much more that what the spec's say?
The 1k2 and the 360R resistors as well as the 1+1k resistors in the voltage divider are simply cut and paste from National datasheet and Application notes. The same values are used in a number of publications. Panson uses 1k+1K in the voltage divider as well.
However, I just now realized that I have put the voltage divider on the wrong side of the RB's.... The voltage divider should sit pre RB's... Could that be the root cause to my problems with the Bias earlier? Seems unlikely. but I'll try it tomorrow.
In my MOSFET version I have put 120R as RB and 2x6k2 as voltage divder - as per National Application Note.
The 100pF G-D cap is put there as per another old MOSFET deisgn, I believe it comes from Elliot Sound Projectcs, but I canät find the reference right now. In what way do you see them as counterproductive?
Drivers for +/-55V supply
What do you think about Sanken 2SA1668and 2SC4382 from Sanken as drivers? 200 VCE, IC 2A. I can get them from LItteldiode.
What do you think about Sanken 2SA1668and 2SC4382 from Sanken as drivers? 200 VCE, IC 2A. I can get them from LItteldiode.
Segran said:
Interesting project 🙂 If the rails are +-55V like in your schematics then I'd expect you to get 100W+ without problems if the circuit is correct. Amps with 60V idle rails tend to produce about 150W in 8 ohms for example.
Please drop me an email: megajocke AT gmail DOT com
Okay, that explains a lot. It seems strange though that you can get that much output. Is it an 8 ohm load? You must have been EXTREMELY lucky with the current gain of those output transistors you have. The LME chip shouldn't reach its current limit in this circuit. Even with the output at 55V and output node at 0V only 20mA will flow from the driver chip.
The 1k2 resistor in series with the whole Vbe multiplier is the reason you can't get the bias down with sensible base resistor values. 2.8mA flows through the bias circuit (see datasheet). This will drop 3.4V over this resistor and then you have at least the vbe of the vbe multiplier - minimum bias voltage of 4V... Way too high! If you short the 1k2 resistor it should work.
It will perform it's feedback-bypasses-output-at-high-frequencies job better if put before the base resistors improving HF stability. But I'd rather not use that kind of circuit at all, but opinions differ - Leach likes that kind of feedback for example.
I seems the application note doesn't want to show when I try looking at it. I'd probably design the network between driver outputs for 10-25mA idle current and put a cap in parallell to help turn off fets.
The thing with the G-D caps is that they increases the current needed for a given slew rate, and you usually want all available current to drive the output FETs. But the caps might be needed for the compensation that is used now.
Thanks megajocke, have changed R13 to 10R and R15/R16 to 100R. That solved the previous problem with the bias. I haven't increased R14 yet.
BUT! Now the output is limited to 40W..... This doesn't make sense to me. Could it be that the increased current in BiasP/BiasM is internally limiting the output from Sink/Source? Is this the Baker clamp that now is working as intended?
If so, I have a choice of finding a combination of resistor values that will give med the 70W I'd like to see, or redesigning the PCB's. This would be rather tricky since there isn't any space to expand the PCB....
BUT! Now the output is limited to 40W..... This doesn't make sense to me. Could it be that the increased current in BiasP/BiasM is internally limiting the output from Sink/Source? Is this the Baker clamp that now is working as intended?
If so, I have a choice of finding a combination of resistor values that will give med the 70W I'd like to see, or redesigning the PCB's. This would be rather tricky since there isn't any space to expand the PCB....
This is from TAIWAN Diyer, hard-wired style....
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I don't think you need to do anything to R14, it's probably OK.
Pretty strange that the output power decreased though. I'd add drivers, performace won't be very good without.
I'd attach them to the heatsink and bend the output transistor base lead out of its hole and connect the driver stage with wires.
Look att coffin's amp, the whole amp is built that way! 😎
Pretty strange that the output power decreased though. I'd add drivers, performace won't be very good without.
I'd attach them to the heatsink and bend the output transistor base lead out of its hole and connect the driver stage with wires.
Look att coffin's amp, the whole amp is built that way! 😎
Those 2SA1668 and 2SC4382 look like they would work but were a bit slow and expensive.
I have some stock of those transistors mentioned earlier as I do repairs. If you want some of the MJE1503x they are 15kr. MJE34x 7kr. FJL 18kr. Frakt & sånt 30kr, moms tillkommer.
I have some stock of those transistors mentioned earlier as I do repairs. If you want some of the MJE1503x they are 15kr. MJE34x 7kr. FJL 18kr. Frakt & sånt 30kr, moms tillkommer.
Sorry, I made a fool of myself -again.. The scope was set to "Uncalibrated". When re-doing the measurement I got 70 Vp-p across a 7R resistive load and 80 V p-p across my 8 ohm speakers (sine wave 1kHz). That's 10 A p-p, 800 W p-p! Not bad for TIP35/36 without drivers...
The measurement was made just at the virge of distortion setting in (yruned out to be the negative oart of the sine wave, indicating that the PNP has lower gain than the NPN - true, they measured 104 v.s 128 in gain.
And yes, the sound is noticeably improved after the changes megajocke suggested. Your where spot on - I'm impressed!
I guess I could fit drivers if I really wanted to, but it would be air-to-air wiring, much like coffin's amp. But I'm satisfied for the time being at least. The tranny in the Marantz isn't going to cope with more power anyway.
Just to show you that the project actually exist and that I really was short of space when designing the PCB's - here's a snap shot of both channels fitted in the Marantz receiver.
Thanks again megajocke for you interest and help, I'll be contacting you about the drivers for the next project - this time I wont take short cuts, I want the best performance possible.
The measurement was made just at the virge of distortion setting in (yruned out to be the negative oart of the sine wave, indicating that the PNP has lower gain than the NPN - true, they measured 104 v.s 128 in gain.
And yes, the sound is noticeably improved after the changes megajocke suggested. Your where spot on - I'm impressed!
I guess I could fit drivers if I really wanted to, but it would be air-to-air wiring, much like coffin's amp. But I'm satisfied for the time being at least. The tranny in the Marantz isn't going to cope with more power anyway.
Just to show you that the project actually exist and that I really was short of space when designing the PCB's - here's a snap shot of both channels fitted in the Marantz receiver.
Thanks again megajocke for you interest and help, I'll be contacting you about the drivers for the next project - this time I wont take short cuts, I want the best performance possible.
Sorry, I made a fool of myself -again.. The scope was set to "Uncalibrated". When re-doing the measurement I got 70 Vp-p across a 7R resistive load and 80 V p-p across my 8 ohm speakers (sine wave 1kHz). That's 10 A p-p, 800 W p-p! Not bad for TIP35/36 without drivers...
The measurement was made just at the virge of distortion setting in (yruned out to be the negative oart of the sine wave, indicating that the PNP has lower gain than the NPN - true, they measured 104 v.s 128 in gain.
And yes, the sound is noticeably improved after the changes megajocke suggested. Your where spot on - I'm impressed!
I guess I could fit drivers if I really wanted to, but it would be air-to-air wiring, much like coffin's amp. But I'm satisfied for the time being at least. The tranny in the Marantz isn't going to cope with more power anyway.
Just to show you that the project actually exist and that I really was short of space when designing the PCB's - here's a snap shot of both channels fitted in the Marantz receiver.
Thanks again megajocke for you interest and help, I'll be contacting you about the drivers for the next project - this time I wont take short cuts, I want the best performance possible.
The measurement was made just at the virge of distortion setting in (yruned out to be the negative oart of the sine wave, indicating that the PNP has lower gain than the NPN - true, they measured 104 v.s 128 in gain.
And yes, the sound is noticeably improved after the changes megajocke suggested. Your where spot on - I'm impressed!
I guess I could fit drivers if I really wanted to, but it would be air-to-air wiring, much like coffin's amp. But I'm satisfied for the time being at least. The tranny in the Marantz isn't going to cope with more power anyway.
Just to show you that the project actually exist and that I really was short of space when designing the PCB's - here's a snap shot of both channels fitted in the Marantz receiver.
Thanks again megajocke for you interest and help, I'll be contacting you about the drivers for the next project - this time I wont take short cuts, I want the best performance possible.
