Hey Kanwar ....
In terms of bias current, it is reasonably stable. The amplifier does take some time to settle when it is first set up. This takes around 5 to 10 minutes depending on the heatsink size. During this time the user needs to keep an eye on the bias current and make adjustments up or down. After the initial setup the amplifier will reach the set bias within a few minutes and maintain it quite well. Naturally if the heatsinks are very hot the bias will back off to a low level, lower than most amps.
In terms of output offset I have not noticed any deviation more than a few millivolts over a wide temperature range. I imagine the negative feedback takes care of that.
In terms of distortion & noise, I must confess that I have not taken measurements over a wide temperature range.
I do not consider these parameters important for the applications this amp was intended for, mainly parties and PA work. It would though make a great subwoofer amp.
In comparison with all the above, the other amp I posted in "Power amp under development" is much better, but not as efficient.
Cheers
Workhorse said:
In terms of bias current, it is reasonably stable. The amplifier does take some time to settle when it is first set up. This takes around 5 to 10 minutes depending on the heatsink size. During this time the user needs to keep an eye on the bias current and make adjustments up or down. After the initial setup the amplifier will reach the set bias within a few minutes and maintain it quite well. Naturally if the heatsinks are very hot the bias will back off to a low level, lower than most amps.
In terms of output offset I have not noticed any deviation more than a few millivolts over a wide temperature range. I imagine the negative feedback takes care of that.
In terms of distortion & noise, I must confess that I have not taken measurements over a wide temperature range.
I do not consider these parameters important for the applications this amp was intended for, mainly parties and PA work. It would though make a great subwoofer amp.
In comparison with all the above, the other amp I posted in "Power amp under development" is much better, but not as efficient.
Cheers
WARNING
There is an error in the schematics for both the low and high power versions of this amp.
The resistor on the emitter of the second constant current source should be 47 ohms not 22 ohms.
This is R17 in both schematics.
Attached is the corrected low power version
There is an error in the schematics for both the low and high power versions of this amp.
The resistor on the emitter of the second constant current source should be 47 ohms not 22 ohms.
This is R17 in both schematics.
Attached is the corrected low power version
And the corrected high power version.
Cheers ......and sorry for any inconvenience (and blown FETS?).
Cheers ......and sorry for any inconvenience (and blown FETS?).
Revised Project
There have been requests (via email mainly) for the high power version to be further developed. Generally the requests are for more power.
The schematic of this amp has been re-done plus new PCB designs to cater for differing needs of DIY'ers.
Here's the latest schematic;
Cheers
There have been requests (via email mainly) for the high power version to be further developed. Generally the requests are for more power.
The schematic of this amp has been re-done plus new PCB designs to cater for differing needs of DIY'ers.
Here's the latest schematic;
Cheers
Input & Driver PCB
This is the driver PCB layout & tracks. Note this PCB has not yet been built so error checking is required.
This is the driver PCB layout & tracks. Note this PCB has not yet been built so error checking is required.
Hi quasi !
I see 2 minor problems with your circuit... As you have cascoded T5/7, their "swing" is limited to ~4v (8.2v - vgs=4.5). But the 1st LTP can output up to 8v, driving T5/7 into heavy reverse bias. This could create problems when amp reaches slewratelimit or clipping. Have you verified that ? A simple solution would be putting 2 antiparalell diodes between the output of the 1st ltp (r4/r9). Also, c8/c9 are connected to points with low voltageswing (~constant voltage from the cascodes), are they still effective enough to keep the amp stable ? You might connect the caps to the drains of T8/9.
BTW, you should also remove/change the comments about T7/9...
Don't be angry with me, but with an amp having that much power you might be more careful.
I still like this design... 😉
Mike
I see 2 minor problems with your circuit... As you have cascoded T5/7, their "swing" is limited to ~4v (8.2v - vgs=4.5). But the 1st LTP can output up to 8v, driving T5/7 into heavy reverse bias. This could create problems when amp reaches slewratelimit or clipping. Have you verified that ? A simple solution would be putting 2 antiparalell diodes between the output of the 1st ltp (r4/r9). Also, c8/c9 are connected to points with low voltageswing (~constant voltage from the cascodes), are they still effective enough to keep the amp stable ? You might connect the caps to the drains of T8/9.
BTW, you should also remove/change the comments about T7/9...
Don't be angry with me, but with an amp having that much power you might be more careful.
I still like this design... 😉
Mike
Me angry?
I'll never get angry at you MikeB, you're always offering good advice.
Shall I just change the value of D4 to 12 or 13 volts?
Yes the comment should read T8 & T9 dissipate about 5 watts total and should be mounted on the heatsink.
With respect to C8 & C9, I have not had any stabilty issues with this arrangement. A similar treatment could be effected to T8 & T9 if there were any problems.
Cheers
I'll never get angry at you MikeB, you're always offering good advice.
Shall I just change the value of D4 to 12 or 13 volts?
Yes the comment should read T8 & T9 dissipate about 5 watts total and should be mounted on the heatsink.
With respect to C8 & C9, I have not had any stabilty issues with this arrangement. A similar treatment could be effected to T8 & T9 if there were any problems.
Cheers
Hi quasi !
Increasing d4 would also do the job, but limit positive swing. I think the diodes would be best choice, with +/- 0.3v the second diffamp should still saturate, but as you have already done the layout...
Mike
Increasing d4 would also do the job, but limit positive swing. I think the diodes would be best choice, with +/- 0.3v the second diffamp should still saturate, but as you have already done the layout...
Mike
Going to be hard to fit the diodes in the PCB.
What about changing R4 & R9 to 2K2 each. This will limit the swing on LTP1 to 4 volts and still provide 2 volts for T6. Then I can leave D4 as 8v2.
What do you think?
What about changing R4 & R9 to 2K2 each. This will limit the swing on LTP1 to 4 volts and still provide 2 volts for T6. Then I can leave D4 as 8v2.
What do you think?
Yes, reducing r4/9 does the job, you should check closely, the 2.2k gives max ~4.4v, i am not sure about the vgs-threshold of the cascode device, i remember 3.6-4.5v for hexfets. 8.2-4.5 will be 3.7v, that would be still enough to cause problems.
2 1n4148 are quite small ?
Mike
2 1n4148 are quite small ?
Mike
Ok..I'll see if I can shift some components around.
Off Topic; Maybe I mentioned it before, I have a young friend near you in Castrop-Rauxal. He lived with us for nearly a year while he went to school here.
Cheers.
Off Topic; Maybe I mentioned it before, I have a young friend near you in Castrop-Rauxal. He lived with us for nearly a year while he went to school here.
Cheers.
Re: Another quasi-complimentary design
I think you should have separate input/VAS positive supply.
Like 50+(15-20) volt regulated.
Gives +65-70 Volt.
Otherwise, I think this amp might work.
quasi said:This one is a variation of a PA amp that I built many years ago. The changes were prompted by learnings made during my post "Power Amp Under Development".
I haven't built it yet. Whadya'll think?
Cheers
I think you should have separate input/VAS positive supply.
Like 50+(15-20) volt regulated.
Gives +65-70 Volt.
Otherwise, I think this amp might work.
An increased +ve voltage for the input and VAS was implemented later in the thread.
Regulating it is a good idea.
Cheers
Regulating it is a good idea.
Cheers
Quasi,
I do like your amp, it has an elegance to it.......
Can you tell me what schematics program you use? It looks terrific!
Cheers,
Hugh
I do like your amp, it has an elegance to it.......
Can you tell me what schematics program you use? It looks terrific!
Cheers,
Hugh
Hi Hugh,
I use;
http://www.abacom-online.de/uk/html/body_splan.html
Do you have any comments / suggestions about the diodes suggested by MikeB or anything else in the cct ?
Cheers
I use;
http://www.abacom-online.de/uk/html/body_splan.html
Do you have any comments / suggestions about the diodes suggested by MikeB or anything else in the cct ?
Cheers
Quasi,
Thanks for the link; I see they have an agent here in Victoria; I'll give 'em a ring. I love the quality of the component graphics, very pretty!
Your amp is within spitting distance of world class, IMHO. I really like the idea of diff drive from the VAS rather than a phase splitter at the output stage. This improves quality all around and enhances rail efficiency.
The big issues with high power quasis are cross conduction, stability with a variety of loads, and asymmetric output impedance from the two halves.
However, a few issues which you might consider (I have no idea if these are valid for your design as I've not built it) - and they are only 1 man's opinion - are:
#1 There are different propagation delays for the two output halves, since one is common source and the other is common drain. This can lead to cross conduction and also hugely affects stability. I'm in some doubt as to whether a cascode is a good idea; it might be better to simply use a slower device for T7 and delete the cascode altogether, thus conferring full signal swing at T7.
#2 To speed up the drive to the lower rail, I'd actually delete the T9 cascode altogether and replace with a resistor/cap parellel combination to drop around 40V. You'll need a high voltage transistor for T5, too, a fast one (100MHz) of reasonable dissipation, up to 5W. This will increase rail swing on T5 and may even reduce lag comp to around 10pF across T5 and T7.
#3 I'm not sure that you'd need diodes across R4/9 as Michael has said, because the transconductance will ensure that each side of the diff pair T5/T7 ever even comes close to 0mA. I doubt the current swing was more than 50%, down to 11mA, up to 33mA max. However, that said, clamp diodes certainly won't do any harm although you'd need to choose low capacitance diodes such as 4148s so as not to interfere with C5/R6, which is pretty finely tuned.
#4 In any quasi-comp, there is huge gain asymmetry between upper and lower halves, because of the additional gain of the common drain lower rail devices. I have not investigated this in theory or practice, but this means feedback factor between output havles will vary hugely. I suspect this has quite an influence on sonics, creating large H2/H3 distortion artefacts. You could tune this OLG asymmetry by varying the resistor in series with T5 collector and R16, since this resistor, with R16, will throw away excessive gain in the lower half. You'd need to be careful, of course, but I suspect this regime would both simplify and enhance the amp's performance, and it may even reduce noise.
Just a few ideas, don't know if they are useful....... I'm sure Kanwar has lots of answers on this, so I'll defer to him.
Cheers,
Hugh
Thanks for the link; I see they have an agent here in Victoria; I'll give 'em a ring. I love the quality of the component graphics, very pretty!
Your amp is within spitting distance of world class, IMHO. I really like the idea of diff drive from the VAS rather than a phase splitter at the output stage. This improves quality all around and enhances rail efficiency.
The big issues with high power quasis are cross conduction, stability with a variety of loads, and asymmetric output impedance from the two halves.
However, a few issues which you might consider (I have no idea if these are valid for your design as I've not built it) - and they are only 1 man's opinion - are:
#1 There are different propagation delays for the two output halves, since one is common source and the other is common drain. This can lead to cross conduction and also hugely affects stability. I'm in some doubt as to whether a cascode is a good idea; it might be better to simply use a slower device for T7 and delete the cascode altogether, thus conferring full signal swing at T7.
#2 To speed up the drive to the lower rail, I'd actually delete the T9 cascode altogether and replace with a resistor/cap parellel combination to drop around 40V. You'll need a high voltage transistor for T5, too, a fast one (100MHz) of reasonable dissipation, up to 5W. This will increase rail swing on T5 and may even reduce lag comp to around 10pF across T5 and T7.
#3 I'm not sure that you'd need diodes across R4/9 as Michael has said, because the transconductance will ensure that each side of the diff pair T5/T7 ever even comes close to 0mA. I doubt the current swing was more than 50%, down to 11mA, up to 33mA max. However, that said, clamp diodes certainly won't do any harm although you'd need to choose low capacitance diodes such as 4148s so as not to interfere with C5/R6, which is pretty finely tuned.
#4 In any quasi-comp, there is huge gain asymmetry between upper and lower halves, because of the additional gain of the common drain lower rail devices. I have not investigated this in theory or practice, but this means feedback factor between output havles will vary hugely. I suspect this has quite an influence on sonics, creating large H2/H3 distortion artefacts. You could tune this OLG asymmetry by varying the resistor in series with T5 collector and R16, since this resistor, with R16, will throw away excessive gain in the lower half. You'd need to be careful, of course, but I suspect this regime would both simplify and enhance the amp's performance, and it may even reduce noise.
Just a few ideas, don't know if they are useful....... I'm sure Kanwar has lots of answers on this, so I'll defer to him.
Cheers,
Hugh
- Status
- Not open for further replies.