If you are going to make a new PCB then you could also separate the decoupling (dirty) and signal (clean) grounds like is usually done.
Do you mean you are going to put in 2 pairs of output devices? It should be possible to use wires there as long as inductance is kept low by keeping the three leads to each transistor close together. IIRC I did use 100mm or so wires in an amp I built once and it worked fine.
Do you mean you are going to put in 2 pairs of output devices? It should be possible to use wires there as long as inductance is kept low by keeping the three leads to each transistor close together. IIRC I did use 100mm or so wires in an amp I built once and it worked fine.
MOSFET version
Here's they MOSFET version same-same, but different... Still not sure about C3/C27, will keep them for the layout.
The same applies for this - additional output stage on a separate PCB connected by wires. Guess the wires will be 100-150 mm max.
I've seen a lot of designs using upto 6 or even 8 pairs for outputs. I'm not going down that route since it's never going to be used for PA or on stage. To spread the outputs over the heatsinks when using 6 pairs they are bound to have quite long wires or tracks on PCB, so I guess the lenght of the wires are not to critical - at least not for PA use....🙂
I will start using a 2x30VAC transformer, as I have it on hand. DC rails of +/-42V will limit the power, but then I can start with only one pair. 😉
Here's they MOSFET version same-same, but different... Still not sure about C3/C27, will keep them for the layout.
The same applies for this - additional output stage on a separate PCB connected by wires. Guess the wires will be 100-150 mm max.
I've seen a lot of designs using upto 6 or even 8 pairs for outputs. I'm not going down that route since it's never going to be used for PA or on stage. To spread the outputs over the heatsinks when using 6 pairs they are bound to have quite long wires or tracks on PCB, so I guess the lenght of the wires are not to critical - at least not for PA use....🙂
I will start using a 2x30VAC transformer, as I have it on hand. DC rails of +/-42V will limit the power, but then I can start with only one pair. 😉
10r for gate resistors may be too low.
D21 & D24 are diodes not Zeners.
Q1 must monitor output temperature.
Q5 is in the wrong place.
Why is R23 not equal to [r24+r25]?
R26 & R28 seem a bit small.
I still don't understand why you need a two stage amplifier.
D21 & D24 are diodes not Zeners.
Q1 must monitor output temperature.
Q5 is in the wrong place.
Why is R23 not equal to [r24+r25]?
R26 & R28 seem a bit small.
I still don't understand why you need a two stage amplifier.
Thanks Andy for your design review. I've never played with MOSFET's before so there is an even steeper learning curve for me than with the BJT's...
"10r for gate resistors may be too low" - you are right, Rg's should probably be somewhere 100-500R. In National AN-1645 they came up with different values on the P-channel compared to the N-channel, so I guess the best would be using 500R trimmer for RG's.
"D21 & D24 are diodes not Zeners" - right again, should be 1N4148, and the zeners should perhaps be 12V instead of 10V.
"Q1 must monitor output temperature"- absolutely, it will be mounted on the heatsink.
"Q5 is in the wrong place" - this puzzles me... The only thing I find wrong here is that D24&24 should be connected between -55V and Gate? Also, shouldn't D21&22 be connected between Gate and the other side of R19?😕
"Why is R23 not equal to [r24+r25]" - No reason, they should be equal. Considering the high gain in the 2 stages you pointed out on the BJT version I will probabaly reduce R23 and [R24+R25] to about 18k. Still don't know if I will keep R23/C34, have to test and see for myself if there is any significant difference.
"R26 & R28 seem a bit small" - this is what P-A Sjöström uses in his Gainclone, and that I have tested. I don't understand the reason for using higher values on this amp?
"I still don't understand why you need a two stage amplifier"- I believe it's because of the design of the DC-servo, causing the input impedance to be rather low (1K), hence the buffer stage.
It would be easier to loose the DC-servo and input stage all together, but going through a number of LME49810 and LM3886 I have noticed that the DC-level on the output varies quite a bit. I'm maybe overly cautious about the DC-level, the detection level of my DC-protection is +/-0,6V....
BTW, since I'm nearing completion of the PCB, here it is for all of you to criticise. Component numbers match the MOSFET schematics but i will use the same PCB for BTJ's as well.
"10r for gate resistors may be too low" - you are right, Rg's should probably be somewhere 100-500R. In National AN-1645 they came up with different values on the P-channel compared to the N-channel, so I guess the best would be using 500R trimmer for RG's.
"D21 & D24 are diodes not Zeners" - right again, should be 1N4148, and the zeners should perhaps be 12V instead of 10V.
"Q1 must monitor output temperature"- absolutely, it will be mounted on the heatsink.
"Q5 is in the wrong place" - this puzzles me... The only thing I find wrong here is that D24&24 should be connected between -55V and Gate? Also, shouldn't D21&22 be connected between Gate and the other side of R19?😕
"Why is R23 not equal to [r24+r25]" - No reason, they should be equal. Considering the high gain in the 2 stages you pointed out on the BJT version I will probabaly reduce R23 and [R24+R25] to about 18k. Still don't know if I will keep R23/C34, have to test and see for myself if there is any significant difference.
"R26 & R28 seem a bit small" - this is what P-A Sjöström uses in his Gainclone, and that I have tested. I don't understand the reason for using higher values on this amp?
"I still don't understand why you need a two stage amplifier"- I believe it's because of the design of the DC-servo, causing the input impedance to be rather low (1K), hence the buffer stage.
It would be easier to loose the DC-servo and input stage all together, but going through a number of LME49810 and LM3886 I have noticed that the DC-level on the output varies quite a bit. I'm maybe overly cautious about the DC-level, the detection level of my DC-protection is +/-0,6V....
BTW, since I'm nearing completion of the PCB, here it is for all of you to criticise. Component numbers match the MOSFET schematics but i will use the same PCB for BTJ's as well.
Hi,Segran said:
gate resistors should preferably be soldered direct to the gate lead using the very shortest resistor lead possible.
The 10V+diode might be too high. Look at the 240 datasheet and see what current passes with Vgs=10.6V.
Hi Andrew,
"gate resistors should preferably be soldered direct to the gate" - the length of the tracks Rg-G is 12,5/11,5mm on the PCB. It will be the same on the extra ouput PCB, since Rg's are placed there. I guess this is short enough? Or how critical is it? What is the drawback in longer leads betwee Rg and G?
"The 10V+diode might be too high" - I see what you mean, to have a more rigid protection scheme I should in fact go down to 8,8V (8v2 + 0,6) or even 8,1V (7V5 + 0,6). I have seen only a few design using this type of protection and they where 10V.
Cheers!
"gate resistors should preferably be soldered direct to the gate" - the length of the tracks Rg-G is 12,5/11,5mm on the PCB. It will be the same on the extra ouput PCB, since Rg's are placed there. I guess this is short enough? Or how critical is it? What is the drawback in longer leads betwee Rg and G?
"The 10V+diode might be too high" - I see what you mean, to have a more rigid protection scheme I should in fact go down to 8,8V (8v2 + 0,6) or even 8,1V (7V5 + 0,6). I have seen only a few design using this type of protection and they where 10V.
Cheers!
Most builders look at the spec for maximum Vgs to prevent destruction of the gate structure.
They then adopt ~12V because that's less than the often specified Vgsmax=15V.
Now you are considering the maximum current that you are prepared to accept.
I have 5v6 + diode on a 4pair Lfet stage.
They then adopt ~12V because that's less than the often specified Vgsmax=15V.
Now you are considering the maximum current that you are prepared to accept.
I have 5v6 + diode on a 4pair Lfet stage.
IRFP240 have Vgsmax of +/-20V, so from only that perspective 10 or even 12V zeners would be OK. At Vgs=6V the Drain current is approx 11,5 A, at Vgs=7V the drain current is 21A, so the choice of zenervoltage will have a great impact on overall performance.
If worst case current draw is approximately three times what the nominal impedance would predict, then you can choose a Vgs protection Zener to suit this worst case audio current.
If your rails are +-50V and the speakers are 8ohms, then the transient current that might be demanded is ~ 40/8/0.35~= 15Apk.
It looks like one pair of 240s would need around 5.6V + diode to give around this peak current. Test and listen and decide whether you want to use 5.6V or go a little higher, but I would suggest that 12V would be inapproriate.
If your rails are +-50V and the speakers are 8ohms, then the transient current that might be demanded is ~ 40/8/0.35~= 15Apk.
It looks like one pair of 240s would need around 5.6V + diode to give around this peak current. Test and listen and decide whether you want to use 5.6V or go a little higher, but I would suggest that 12V would be inapproriate.
Hi Andrew, can you please clarify this sentence:
"Q5 is in the wrong place" , I can't see what's wrong with it....
"Q5 is in the wrong place" , I can't see what's wrong with it....
neither can I.Segran said:Hi Andrew, can you please clarify this sentence:
"Q5 is in the wrong place" , I can't see what's wrong with it....
Must have had a senior moment.
Something is clearly wrong with the connections of the P-channel output to the driver stage. Gate is connected to the wrong side of driver transistor.
I'd connect the zeners before the driver stage instead, where they are now they will not stand a chance against the MJE1503x:s 😛
I'd connect the zeners before the driver stage instead, where they are now they will not stand a chance against the MJE1503x:s 😛
Hi,
"Gate is connected to the wrong side of driver transistor."
Are you referring to the layout? Collector of Q3 (pin 2) connected via R33 to Gate of Q5 (Pin 1)?
Maybe it's 'cause it's kinda late, or I am just thick headed, but I can't see whats wrong. 😕
Or are you talking about the schematics? AndrewT mentioned "Q5 is in wrong place", I didn't gety that either and when checking with Andrew again he blamed a "senior moment"..
I'm totally lost...
"Gate is connected to the wrong side of driver transistor."
Are you referring to the layout? Collector of Q3 (pin 2) connected via R33 to Gate of Q5 (Pin 1)?
Maybe it's 'cause it's kinda late, or I am just thick headed, but I can't see whats wrong. 😕
Or are you talking about the schematics? AndrewT mentioned "Q5 is in wrong place", I didn't gety that either and when checking with Andrew again he blamed a "senior moment"..
I'm totally lost...
I can see again.
the pnp driver should be connected to the -ve supply as a mirror image to the npn driver.
The Pfet gate should connect to the emitter side of the driver.
the pnp driver should be connected to the -ve supply as a mirror image to the npn driver.
The Pfet gate should connect to the emitter side of the driver.
Here is a corrected (not necessarily correct)schematics for my MOSFET version. I have also reduced the gain in stage 2 and moved the Vgs protection to the driver Base connections as suggested by megajocke.
I wonder if I need to replace R15/16 with 0R/jumpers or what.
Another question is if I should split R34 in half and connect them to R19/20 (output)...
Values for R9/R33 will be decided during testing - if it works at all...
I wonder if I need to replace R15/16 with 0R/jumpers or what.
Another question is if I should split R34 in half and connect them to R19/20 (output)...
Values for R9/R33 will be decided during testing - if it works at all...
Here is a correct schematics for my MOSFET version. I have also reduced the gain in stage 2 and moved the Vgs protection to the driver Base connections as suggested by megajocke.
I wonder if I need to replace R15/16 with 0R/jumpers or what.
Another question is if I should split R34 in half and connect them to R19/20 (output)...
Values for R9/R33 will be decided during testing - if it works at all...
I wonder if I need to replace R15/16 with 0R/jumpers or what.
Another question is if I should split R34 in half and connect them to R19/20 (output)...
Values for R9/R33 will be decided during testing - if it works at all...
The two channels Power Amplifier using LME4910 driver and MOTOROLA MJ15003/MJ15004 Power Transistors.
The schematic of amplifier PCB is a little similar the NS LME49810TB datasheet FIGURE 1.
The voltage supply LME49810 driver is about 72Vdc, and 60Vdc for power transistors.
There are three pears power transistor for each channel.
The schematic of amplifier PCB is a little similar the NS LME49810TB datasheet FIGURE 1.
The voltage supply LME49810 driver is about 72Vdc, and 60Vdc for power transistors.
An externally hosted image should be here but it was not working when we last tested it.
There are three pears power transistor for each channel.
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.
