a quick sim of a "half cell" suggests a fair amount more gain can be had by using a cfp for your follower - mosfet bias permitting, but probably hopelessly complcates overload behavior
Hi jcx,
A CFP instead of the BC546C? The gain's huge already, believe me.
" probably hopelessly complicates overload behaviour" - as it is I've had to put in a clamp diode to stop the base collector forward biassing and switching on the MOSFETs at clipping on the off cycle.
Cheers,
greg
A CFP instead of the BC546C? The gain's huge already, believe me.
" probably hopelessly complicates overload behaviour" - as it is I've had to put in a clamp diode to stop the base collector forward biassing and switching on the MOSFETs at clipping on the off cycle.
Cheers,
greg
Just as a matter of interest I decided to sim my design with a Leach cascode front end (pictured below) to check for PSRR. This can be compared with my cascode front end at post 40 -
http://www.diyaudio.com/forums/showthread.php?postid=786965#post786965
Cheers,
Greg
http://www.diyaudio.com/forums/showthread.php?postid=786965#post786965
Cheers,
Greg
Attachments
Well Hi all,
Between festivities and libations I decided to build a new piece of test equipment to test my amps - a 10KHz tailored impedance square wave generator. I needed one which had sufficiently fast rise time and worked into a capacitative load, so I could bypass the Rinput filter and test closed loop character. A very fast rise time better stimulates the UHF character of an amplifier.
Spec : 10KHz, rise time < 400nS , fall time < 100nS. Zout 47R //1000pF.
Here is the GB300D with it's output . no 160KHz input filter.
Cheers,
Greg
Between festivities and libations I decided to build a new piece of test equipment to test my amps - a 10KHz tailored impedance square wave generator. I needed one which had sufficiently fast rise time and worked into a capacitative load, so I could bypass the Rinput filter and test closed loop character. A very fast rise time better stimulates the UHF character of an amplifier.
Spec : 10KHz, rise time < 400nS , fall time < 100nS. Zout 47R //1000pF.
Here is the GB300D with it's output . no 160KHz input filter.
Cheers,
Greg
Attachments
I think not mastertech,
Interestingly the rise and fall times are virtually those of the square wave generator... so I guess I'll need to do an even faster one.
That's the trouble with building really fast amplifiers, in this case full gain bandwidth >5MHz, is that your equipment has to be upgraded as well.
Perhaps I should stick to ordinary amps in pretty boxes and high bias reviews. Real challenging stuff. Then find some marketing person to shout how wonderful it is - without even hearing it ???
Cheers,
Greg
Interestingly the rise and fall times are virtually those of the square wave generator... so I guess I'll need to do an even faster one.
That's the trouble with building really fast amplifiers, in this case full gain bandwidth >5MHz, is that your equipment has to be upgraded as well.
Perhaps I should stick to ordinary amps in pretty boxes and high bias reviews. Real challenging stuff. Then find some marketing person to shout how wonderful it is - without even hearing it ???
Cheers,
Greg
Home sweet home
What is the V/DIV on your picture Greg? Without it seems difficult to ***** the result!
5V/div??
\Jens
What is the V/DIV on your picture Greg? Without it seems difficult to ***** the result!
5V/div??
\Jens
Hi Jens,
When I cropped the photo I left in the V/div selector. 1V /div.
It's roughly 1W. The sq wave gen's just a lashup at present. Needs a box, connectors and a pot for functionality.
Cheers,
greg
When I cropped the photo I left in the V/div selector. 1V /div.
It's roughly 1W. The sq wave gen's just a lashup at present. Needs a box, connectors and a pot for functionality.
Cheers,
greg
Jens, you can actually see the knob at 1 V/div. Check the black stripe on the knob.JensRasmussen said:
Greg,
amplifierguru said:
I know, I was unsure about the x5 botton......
I often use a x10 probe, I thought that you might do the same
How about a rail to rail pic?
\Jens
Hi Jens,
There's no way I would do a rail to rail at 10K sq wave with 3MHz BW.
It's inviting disaster, irrelevant and stoopid. Such an amp to handle this could easily be built but would add complexity unwarranted for audio. But you knew that.
The max I would test , bandlimited, at 10KHz is about 50W for this size of amplifier. I'm happy with that as it meets practical requirement.
Perhaps I've been doing this too long?
Cheers,
Greg
There's no way I would do a rail to rail at 10K sq wave with 3MHz BW.
It's inviting disaster, irrelevant and stoopid. Such an amp to handle this could easily be built but would add complexity unwarranted for audio. But you knew that.
The max I would test , bandlimited, at 10KHz is about 50W for this size of amplifier. I'm happy with that as it meets practical requirement.
Perhaps I've been doing this too long?
Cheers,
Greg
I think our Jens is trying to tell you something Greg,dont you think
Jens is a real business man, no doubt
Jens is a real business man, no doubt
I don't know if I'm allowed to post on this thread - I might be deemed to be promoting my own design.
Hi Mastertech,
Too subtle for moi, I'm afraid. I tend not to be.😀
Cheers,
Greg
Hi Mastertech,
Too subtle for moi, I'm afraid. I tend not to be.😀
Cheers,
Greg
amplifierguru said:
Greg,
Do you think it's a good idea to sell an amp that is not uconditionally
stable into any C load?
I'm thinking of all those alphacore foil speaker cables out there
which represent a significant C load with very low series R and L.
Cheers,
Terry
Hi Terry,
Don't know of these cables. The post you quote was me indicating at that point in time that this larger amp may need an output R//L for capacitative load isolation.
The pic you didn't include showed typical controlled ringing with capacitative loads , quite acceptable behaviour. But there are always component spreads and external DIY vagaries which may lead to parasitics. These you can't know until they occur.
Cheers,
Greg.
Don't know of these cables. The post you quote was me indicating at that point in time that this larger amp may need an output R//L for capacitative load isolation.
The pic you didn't include showed typical controlled ringing with capacitative loads , quite acceptable behaviour. But there are always component spreads and external DIY vagaries which may lead to parasitics. These you can't know until they occur.
Cheers,
Greg.
amplifierguru said:
Greg,
And when they occur, it's often at the least opportune moment!
Goertz (alphacore) or the equivalent home made cables are
basically two wide / thin strips of foil placed very closely together
with a thin insulator between them. The geometry varies but the
end result is relatively high C (forcables) and very low series L and
R.
Some amp OP stages that are perfectly happy with 1uF
strapped across them can oscillate driving these cables. So it
is good practice to make the amp unconditionally stable into
all C loads.
The amp can break into a medium level oscillation that it's fuses can
sustain but the speakers can't. I've had this exact problem occur
with a big SS power amp driving some very expensive Goodmans
Axiom 80 drivers through long alphacore cables.
Luckily the drivers were OK.
Cheers,
Terry
Hi Terry,
I appreciate what you're saying. I actually designed in a space for an 'optional' output R//L just in case it was borderline stability.
However, it's a whole lot better than that especially with the input filter in and 100mm of output cable to the speaker terminals. But just possibly, if I had those speaker cables/capacitors, I would be winding 15 turns on a IW resistor.
There are two output takeoffs - before, and after the R//L.
Can't do much more.
Cheers,
Greg
I appreciate what you're saying. I actually designed in a space for an 'optional' output R//L just in case it was borderline stability.
However, it's a whole lot better than that especially with the input filter in and 100mm of output cable to the speaker terminals. But just possibly, if I had those speaker cables/capacitors, I would be winding 15 turns on a IW resistor.
There are two output takeoffs - before, and after the R//L.
Can't do much more.
Cheers,
Greg
- Status
- Not open for further replies.