First of all, hi to everybody, I'm new to the forums! 
I'm fairly young, and hope that you'll be forgiving when I mistype and/or ask stupid questions.
So, I've decided to build myself an solidstate poweramp for use with my guitar. Since this is a guitaramp, there is no need for extreme fidelity, but what I really need (and miss with my other crappy amp) is good SNR. I'm aiming at approx 30 watts into 8 ohms. Since my guitar uses active pickups I'll skip a preamp and design the PA first and let a multieffect box and some equalizers do that job in the meantime.
I'm thinking differential input with CSS, one stage of voltage gain and a MOSFET output stage biased into AB or perhaps pure A. Pretty straight forward, I guess. Somewhat in the lines of this:
http://www.redcircuits.com/Page100.htm
But, some questions:
1 - I would need a temperature stabilizer wouldn't I ?
Whats the best approach? Vbe multiplier where the diodes and pot is now?
2 - And is the configuration of Q7 a good way of driving MOSFETS?
3 - Most of these parts aren't readily avalible for me, so I was thinking of replacing Q1/2 with simple BC546A,
Q7 with a BC639, Q8/9 with a double pair of IRF(9)540N, and the CSS bjts with generic BC546/7
I know this will have some impact on sound quality, but it will still function properly/satisfactory in a guitar amp?
Low gain is better than high gain in most of these BJTs in order to keep feedback fairly low (to minimise danger of instability) right ? I havent quite decided the gain yet, will have to do some measuring first.
Thanks a lot for all the help you are able to give me, I value it alot!
I'm fairly young, and hope that you'll be forgiving when I mistype and/or ask stupid questions.
So, I've decided to build myself an solidstate poweramp for use with my guitar. Since this is a guitaramp, there is no need for extreme fidelity, but what I really need (and miss with my other crappy amp) is good SNR. I'm aiming at approx 30 watts into 8 ohms. Since my guitar uses active pickups I'll skip a preamp and design the PA first and let a multieffect box and some equalizers do that job in the meantime.
I'm thinking differential input with CSS, one stage of voltage gain and a MOSFET output stage biased into AB or perhaps pure A. Pretty straight forward, I guess. Somewhat in the lines of this:
http://www.redcircuits.com/Page100.htm
But, some questions:
1 - I would need a temperature stabilizer wouldn't I ?
Whats the best approach? Vbe multiplier where the diodes and pot is now?
2 - And is the configuration of Q7 a good way of driving MOSFETS?
3 - Most of these parts aren't readily avalible for me, so I was thinking of replacing Q1/2 with simple BC546A,
Q7 with a BC639, Q8/9 with a double pair of IRF(9)540N, and the CSS bjts with generic BC546/7
I know this will have some impact on sound quality, but it will still function properly/satisfactory in a guitar amp?
Low gain is better than high gain in most of these BJTs in order to keep feedback fairly low (to minimise danger of instability) right ? I havent quite decided the gain yet, will have to do some measuring first.
Thanks a lot for all the help you are able to give me, I value it alot!
Welcome to the forum!
As regards your questions 1 and 3... If diodes D1 through D3 are mounted onto the heatsink, they act as thermal stabilizers.
I'd prefer BD139 for Q7. The rest of your replacements are fine by me.
Now over to the gurus of vbe multipliers ... here's a chance to show your stuff.
Regards,
Milan
As regards your questions 1 and 3... If diodes D1 through D3 are mounted onto the heatsink, they act as thermal stabilizers.
I'd prefer BD139 for Q7. The rest of your replacements are fine by me.
Now over to the gurus of vbe multipliers ... here's a chance to show your stuff.
Regards,
Milan
Hi Northwave,
Good idea you have. Small enough to be successful, and with your requirements laid out.
Just give it a big heatsink so you can play for hours. As Milan pointed out, the bias stage may need a rework, but that can be done later. The first step is to get it working.
Welcome to Diy Audio!
-Chris
Good idea you have. Small enough to be successful, and with your requirements laid out.
Just give it a big heatsink so you can play for hours. As Milan pointed out, the bias stage may need a rework, but that can be done later. The first step is to get it working.
Welcome to Diy Audio!
-Chris
Consider IRF640 and 9540 as complementaries (yes, I know, the datasheet suggests otherwise... until you check the actual numbers in it!). Check that neither part has an N at the end (so, IRF640, NOT IRF640N). This is not as good as IRFP240 and IRFP9140 - these two come as close to complementary as you can get within he IRF lineup.
The 3 diodes and pot should really be replaced by a Vgs multiplier. This would go kind of like this: instead of the diodes and the trim pot, use a small MOSFET (say IRF510, 520, 610), connect drain to the place the top diode anode was, and the source to the place where the bottom end of the pot was. Now, take a 5.6k resistor and connect between gate and drain of this MOSFET. Then, take a trim pot of 10k and connect it's ends between gate and source of this MOSFET. Finally, take a 5.6k resistor and connect between source of MOSFET and wiper of trim pot. Mount MOSFET on the same heatsink with the output MOSFETs.
Regarding your supstitutions, use BD139/140 for Q6 and Q7 if you must, although they will be at the very edge of their maximum Vce at +-40V rails. These will dissipate quite a bit of heat, about 0.6W, which would be too high for a BC639. If you use two pairs of output transistors, consider increasing the source resistors, to perhaps 0.5 or 0.68 ohms. This will produce better current sharing, and also equalize the noncomplementarity somewhat, while providing the same rail loss as the originals. IMHO you should really try to get the IRFP parts, they are MUCH more robust, and designed for much higher power dissipation.
Finally, it may be necessary to have a small cap (say 22-47p) at hand, to connect between B and C of Q7, if the amp oscillates.
The 3 diodes and pot should really be replaced by a Vgs multiplier. This would go kind of like this: instead of the diodes and the trim pot, use a small MOSFET (say IRF510, 520, 610), connect drain to the place the top diode anode was, and the source to the place where the bottom end of the pot was. Now, take a 5.6k resistor and connect between gate and drain of this MOSFET. Then, take a trim pot of 10k and connect it's ends between gate and source of this MOSFET. Finally, take a 5.6k resistor and connect between source of MOSFET and wiper of trim pot. Mount MOSFET on the same heatsink with the output MOSFETs.
Regarding your supstitutions, use BD139/140 for Q6 and Q7 if you must, although they will be at the very edge of their maximum Vce at +-40V rails. These will dissipate quite a bit of heat, about 0.6W, which would be too high for a BC639. If you use two pairs of output transistors, consider increasing the source resistors, to perhaps 0.5 or 0.68 ohms. This will produce better current sharing, and also equalize the noncomplementarity somewhat, while providing the same rail loss as the originals. IMHO you should really try to get the IRFP parts, they are MUCH more robust, and designed for much higher power dissipation.
Finally, it may be necessary to have a small cap (say 22-47p) at hand, to connect between B and C of Q7, if the amp oscillates.
Since there's different packages and no upgrade would be possible with TO220, I've decided to go for the IRFP combo afterall.
Vbe multiplier will be implemented, and increased the source resistors to 0,68 ohms per FET. Swapped Q6/Q7 for your suggestion, didn't consider the dissipation in the driverstage at first.
To the experienced people here:
Gain is decided by R6/R8 right?
What does C5/R9 do ?
And the output filter R16,17/C9?
Thanks alot so far, you've helped me alot!
Parts are now ordered The waiting has started.
Vbe multiplier will be implemented, and increased the source resistors to 0,68 ohms per FET. Swapped Q6/Q7 for your suggestion, didn't consider the dissipation in the driverstage at first.
To the experienced people here:
Gain is decided by R6/R8 right?
What does C5/R9 do ?
And the output filter R16,17/C9?
Thanks alot so far, you've helped me alot!
Parts are now ordered
The waiting has started.Hi,
R9 & C5 are used to shelve off the HF response without introducing a two pole slope as the gain gets to 0db.
Using this method of obtaining HF stability removes or reduces the need for a Miller Compensation cap around the VAS transistor.
This is reputed to sound nicer than Miller comp, but is more difficult to ensure adequate phase margin for all component tolerances. You should probably use an oscilloscope to check stability into a variety of reactive loads.
Your gain setting is correct.
output filter is actually a HF load to help ensure stability.
C7 forming the input filter and DC block seems to be tiny (22mS), but may suit guitar (not bass).
R9 & C5 are used to shelve off the HF response without introducing a two pole slope as the gain gets to 0db.
Using this method of obtaining HF stability removes or reduces the need for a Miller Compensation cap around the VAS transistor.
This is reputed to sound nicer than Miller comp, but is more difficult to ensure adequate phase margin for all component tolerances. You should probably use an oscilloscope to check stability into a variety of reactive loads.
Your gain setting is correct.
output filter is actually a HF load to help ensure stability.
C7 forming the input filter and DC block seems to be tiny (22mS), but may suit guitar (not bass).
if you aim for max
... 30 W RMS into 8 ohm
( Max peak voltage for this power is: 21.91 Volt )
#####################
Transformer:
You should definitively try to lower voltage supply
from 2x40V ... to something like 2x30.
Will give those HEXFET a margin of about 30-21.91 = 8.09 Volt to live within at max output. )
A transformer 2x22VAC 100 VA per channel would be right, for me.
#####################
Q1, Q2, Q4:
As we lower total voltage from 2x40 to 2x30
= 60 Volt
For input, BC546B ( 65 volt ) would be perfect to use for Q1,Q2, Q4 without any overloading problems.
Same for Q3, Q5 in current sources, you can keep using BC546B.
If you can find BC546C, you should buy those instead.
The higher the gain of small signal transistors .. the better!
But otherwise BC546B will work just as good.
#####################
Q6, Q7:
I agree with that BD139 and BD140 are considerably better for Q6, Q7.
As you have lowered voltage to just above 60 volt we can safely use BD139-140 ( max 80 volt ).
( MJE340 MJE350 is only to be a choice
if working with higher voltage
than BD139, BD140 can take. More than 80 Volt.
I have tried making comparing simulation using BD139 and MJE340 in my amplifiers.
And always BD139 shows clearly the better performance figures. )
#####################
For the output:
Even if here is only 30 Watt max I really recommend you use
one robust and heavy duty output pair.
And avoid the trouble with paralleling several output pair.
At max 30 watt at 8 ohm in class AB, we have neither VERY high AVERAGE powers or currents.
One higher power pair is good!
One solid complementary pair
in some of following CAPSULES, CASES will be good:
TO247, TO-3P, TO3, TO264.
#####################
Heatsink.
Using a bigger transistor variant ( 200 degrees max internal temperature ) will also reduce the heatsink size needed.
But anyway I advice you in this case,
if you use those 3 diodes close to heatsink for temp sensing and control
you should NOT have TOO SMALL heatsink,
because this will make higher temperature and lot more temp variation span
and this is not good for any amplifier output stage!
You do not need buy a gigantic heatsink.
Something in the medium size of:
width: 150-200 millimeter ( 6-8 inches )
height: 75-100 millimeter ( 3-4 inches )
#####################
lineup
... 30 W RMS into 8 ohm
( Max peak voltage for this power is: 21.91 Volt )
#####################
Transformer:
You should definitively try to lower voltage supply
from 2x40V ... to something like 2x30.
Will give those HEXFET a margin of about 30-21.91 = 8.09 Volt to live within at max output. )
A transformer 2x22VAC 100 VA per channel would be right, for me.
#####################
Q1, Q2, Q4:
As we lower total voltage from 2x40 to 2x30
= 60 Volt
For input, BC546B ( 65 volt ) would be perfect to use for Q1,Q2, Q4 without any overloading problems.
Same for Q3, Q5 in current sources, you can keep using BC546B.
If you can find BC546C, you should buy those instead.
The higher the gain of small signal transistors .. the better!
But otherwise BC546B will work just as good.
#####################
Q6, Q7:
I agree with that BD139 and BD140 are considerably better for Q6, Q7.
As you have lowered voltage to just above 60 volt we can safely use BD139-140 ( max 80 volt ).
( MJE340 MJE350 is only to be a choice
if working with higher voltage
than BD139, BD140 can take. More than 80 Volt.
I have tried making comparing simulation using BD139 and MJE340 in my amplifiers.
And always BD139 shows clearly the better performance figures. )
#####################
For the output:
Even if here is only 30 Watt max I really recommend you use
one robust and heavy duty output pair.
And avoid the trouble with paralleling several output pair.
At max 30 watt at 8 ohm in class AB, we have neither VERY high AVERAGE powers or currents.
One higher power pair is good!
One solid complementary pair
in some of following CAPSULES, CASES will be good:
TO247, TO-3P, TO3, TO264.
#####################
Heatsink.
Using a bigger transistor variant ( 200 degrees max internal temperature ) will also reduce the heatsink size needed.
But anyway I advice you in this case,
if you use those 3 diodes close to heatsink for temp sensing and control
you should NOT have TOO SMALL heatsink,
because this will make higher temperature and lot more temp variation span
and this is not good for any amplifier output stage!
You do not need buy a gigantic heatsink.
Something in the medium size of:
width: 150-200 millimeter ( 6-8 inches )
height: 75-100 millimeter ( 3-4 inches )
#####################
lineup
AndrewT:
As I thought. Perhaps I'm not as bad at this as I feared.
lineup:
I've decided to go all the way after all, so the amplifier is nearly at stock specs now. The only thing that I've swapped is the small signal transistors in the current sources and input. Actually bought BC546A which I kinda regret now, but it's a minor cost. I'll probably use them for testing, and swap them for something when I decide to order from futurlec.
Transformer is 2x28 160VA, so that's probably between 35-40 volts DC. But couldn't I use a small resistor in series with each supply to the input stage to drop voltage a bit? To ensure that the btjs are operating within limits, I mean. Think I've seen this been somewhere done before.
I've ordered 2 pairs of IRFP complimetary in TO247 casing, so I'll probably do a layout for two pairs. Might just use one pair tho, if I feel that it's sufficient. Heatsink is 160x100mm, so I hope your assumptions are valid.
As I thought. Perhaps I'm not as bad at this as I feared.
lineup:
I've decided to go all the way after all, so the amplifier is nearly at stock specs now. The only thing that I've swapped is the small signal transistors in the current sources and input. Actually bought BC546A which I kinda regret now, but it's a minor cost. I'll probably use them for testing, and swap them for something when I decide to order from futurlec.
Transformer is 2x28 160VA, so that's probably between 35-40 volts DC. But couldn't I use a small resistor in series with each supply to the input stage to drop voltage a bit? To ensure that the btjs are operating within limits, I mean. Think I've seen this been somewhere done before.
I've ordered 2 pairs of IRFP complimetary in TO247 casing, so I'll probably do a layout for two pairs. Might just use one pair tho, if I feel that it's sufficient. Heatsink is 160x100mm, so I hope your assumptions are valid.
Northwave said:
Transformer is 2x28 160VA, so that's probably between 35-40 volts DC.
But couldn't I use a small resistor in series with each supply to the input stage to drop voltage a bit?
To ensure that the btjs are operating within limits, I mean. Think I've seen this been somewhere done before.
I've ordered 2 pairs of IRFP complimetary in TO247 casing, so I'll probably do a layout for two pairs. Might just use one pair tho, if I feel that it's sufficient. Heatsink is 160x100mm, so I hope your assumptions are valid.
2x28VAC will give something like 2x36-41 Volt I guess.
And so a total working voltage of 72-82 volt.
The lower figure is my guess, when playing at MAX 60 watt output.
As a transformer is loaded to give higher currents
then the voltage, that transformer can give, will drop.
It may go down to 2x36 VDC.
At normal lower power output, it may be around 2x40 Volt DC.
Now Q3 Q5 is no problem, they work always at a few volts only.
Q1, Q2, Q4 is also no problem.
As long as amplifier is working as it should
they will work at worst case only 50% of 82 Volt +-a few volts
which will be something around max 45 volts.
So you could safely use BC546B for Q1,Q2,Q4, too!
#####################################
For Q6, Q7 you should use transistors rated at least 100 Volt,
which is 82 Volt + 18Volts safety margin.
If you want to try something else than MJE340, MJE350 for this Q6,Q7 pair,
you may try MJE15030-15031 or ( MJE15028-15029 ).
They are 150V ( 120Volts ) TO220 driver transistors,
which are rather fast. But not as fast as MJE340.
I don't really think they in this case would be better than MJE340-350.
But to know for sure we would have to make a test.
Personally, I don't think MJE15030 would perform much worse than MJE340, though.
MJE15030 - MJE15031 ( MJE15028 - MJE15029 )
is simply one of the best driver pair transistors in the world,
such we could easily get .
For rather fast and linear 150 Volt ( 120 Volt ) audio amplifier operations
at current with a level of something like 50-250 mA
lineup
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.