Yes. Turns out easier that way.
I change the time step depending on what I'm doing. If it's working well and I want a better thd result, I increase the maxstep where it's calculated with len/2**xx, where xx is the power of 2. For quicker sims where finer thd results and fft aren't needed, I just leave that usually at 2**10 or so, and the FFT at 2**10 or even 2**8 if it's too slow and I don't want to wait.
When things work nice and I need a better thd reading, and I want to look at the FFT, I increase the maxstep factor with 2**16, and sometimes FFT at 2**12. With 2**16, we get 64k steps (65536). Anything beyond that isn't really useful and it start to get very very slow.
Yes, but it's general ltspice stuff and not audio specific. I occasionally post there too, and read the posts most of the time as well, depending on the subject. Lots going on, so too much for me to keep up with it all.
Yes, there are no 3055s in cordell's lib, and we're using the mje15030/1 which aren't in his lib either.
And his BC550/60 are the C grade gain types, so that's why it can be a bit confusing to name the devices with a simple C suffix, perhaps it would be better with C_C, that way we know it's both C grade and cordell's.
Besides, if devices were named that way in his lib, we could just point to them with the ako: and have the real part names in the schematics. Plus it would be easy to swap models when needed without altering the schematic all over the place, just with a simple and quick change in the ako: which could be made even faster with several lines and we just rem out all except the one needed.
Sure, why not? I'll have to try that too.
I run sims on several different computers so I like to be able to do a straight install
of LTspice and then add model files as needed unmodified if possible.
Bob C. has MJE15032C/33C there is no reason at all to use the Onsemi models.
Here is a list: CordellAudio.com - SPICE Models
I don't know if you've ever read a book on programming style but making things as
clear as possible, readable, makes it easier for other people to look at your work.
ako's could easily lead to confusion, maybe just for the outputs but I would keep it
to a minimum.
You have a nice style with your schematic drawing I have to say, many professionals
do not.
of LTspice and then add model files as needed unmodified if possible.
Bob C. has MJE15032C/33C there is no reason at all to use the Onsemi models.
Here is a list: CordellAudio.com - SPICE Models
I don't know if you've ever read a book on programming style but making things as
clear as possible, readable, makes it easier for other people to look at your work.
ako's could easily lead to confusion, maybe just for the outputs but I would keep it
to a minimum.
You have a nice style with your schematic drawing I have to say, many professionals
do not.
I know what you mean. I was doing stuff on a laptop and an imac some time ago, and I was using some sync software to keep many things synced up between them two. When the laptop became much less usable (no more battery and all that), I was again confined to a single machine.
But still, when I got tired of having to duplicate models all the time with every new sim, making new folders with all the bits and pieces, I ended up adding certain models to my ltspice lib, to reduce the fooling around with those models.
Since I'm sharing quite a bit, I try to duplicate again, but it gets messy.
I know that list and I use that lib file as well.
And there are differences between those 32/33 and the 30/31. The sims I've been running recently are aimed at the 30/31 pair and not the 32/33. Using those different models would be less like the final design.
I'm not sure if the 30/31 models that came with leach's sim were intact from onsemi, but I have done a tiny bit of tweaking on them, and they're usually not the ones causing issues.
And since Bob's lib isn't complete enough for all our use, we're bound to need other models at some point.
This superleach sim aimed at the 3055s uses models from Bob's lib, the mje15030/31 are from leach's sim lib, and then what's left are the 3055/2955, from various and often dubious origins.
I've done quite a few sims based on the 3055, and I'd like to eventually be able to tweak the models to have at least a good one. In many sims, they do work fine though. It's been with the leach amp sim that I've had the most troubles.
I haven't read a lot of books of that nature, but I've done a fair amount of programming over the years. I know it's a must to properly document and make it clear, so someone else can pick up behind others, and often even ourselves when some time has passed and we forgot completely how we did it before.
That happened to me several times, when having to get back into stuff I wrote long ago. No amount of documentation added would ever be enough to easily get back into something old like that. Tricky thing to do.
It may be so but they also make it easier in some other ways. As long as it's not done for all models in a sim, only for one or two, it ain't too bad.
Well, thanks, but I'm doing this for myself just as much, as I want it as clear as possible for my own understanding.
With spice sims, I've seen so many out there that make advanced stuff and yet so cloudy and messy... I just do what I can to make mine cleaner and easier to read.
However, I avoid having the same circuit in more than one sim with different configuration for various sim types, this is too messy and I couldn't keep up. So I prefer one single sim file, with all the needed sims prepared right into it. So all there is to do is just uncomment what's needed for a sim and the rest is commented for other uses.
Obviously when the circuits get more advanced and we need to run more varied sim types, it can get crowded, and it's not so easy to keep things tidy, but it's better than having a bunch of separate sims.
I tried the trick to move the feedback pickup to the vas and I totally uncoupled the rest after the vas, so nothing could come back from the output stages.
I only rerouted and disconnected, without changing values, and ran the same sim as earlier, so the gain isn't the same and the same sim gives a lower signal level output, taken from the vas output.
This was on that same sim as earlier, the 4 output sets version with tmc instead of the leach compensation. Plus the vas limiter remains disconnected.
And there we go, nothing ugly to report. It all looks good everywhere, right from the first ltp transistor, both currents and voltages look like clean sines from the ltp all the way to the vas output.
Those big spikes must be generated elsewhere, or whatever the cause is, it passes its effects on when the whole circuit is connected.
I didn't try keeping the predrivers connected to the vas and disconnect the rest of the output stage, that's rather complex there and I'm not sure we can really do this, because of the stacking and the setup that takes the output signal to bias the stacks.
Leach’s first version used three Si diodes, (P38 Fig 5 Feb’76 ).
Now that I can see your VI limiter component values, the limiter locus can be computed using Kiwanuka’s analysis.
Vce Ic
0.00000 4.62896
14.98288 3.74449
119.57601 1.92723
Assuming Vbe = 480mV.
Vcc = 60V
If Vbe = 680mV, locus shifts up by about 2A.
Does this agree with your calculations?
Now that I can see your VI limiter component values, the limiter locus can be computed using Kiwanuka’s analysis.
Vce Ic
0.00000 4.62896
14.98288 3.74449
119.57601 1.92723
Assuming Vbe = 480mV.
Vcc = 60V
If Vbe = 680mV, locus shifts up by about 2A.
Does this agree with your calculations?
The superamp right? And are you referring to the bias spreader diodes?
If you're looking at the latest sim for a modified superamp with 4 sets of 3055/2955 outputs, you've noticed the whole protection has changed.
On this latest version being tried, the protections haven't been tested yet, as there are other issues to deal with first.
Before that, on other versions, there was too much oscillation going on to properly test this.
However, on other sims with the same protection setup on a very similar vas and driver topo, what I found is that the sims reveal the results don't match the calculations. Especially when applying a full short, or very hard load.
The protection's action isn't doing a deep enough foldback to keep within the soa, and the protections don't act sharply enough, so they start acting much too early, intruding on legit signal, raising distortion much too early, and then when they should be acting hard to keep soa safe, they're not effective enough and the soa is still violated.
So far I haven't seen this to work "as advertised". It may all look good on paper and in calculations, and they seem to work fine when testing, at first, until more adverse situations are imposed, which should be handled by that protection, and isn't sufficiently handled.
What I find is that because of this insufficient effectiveness and much too early action, more output pairs need to be added to compensate, giving far more soa than really should be needed, just so the protections can work and not be intrusive.
The thing is, unless something is changed in the protection circuit and probably whatever they act on, they are not going to act according to what was calculated.
If someone has a sim with this working properly, I really want to see this.
Definitely, quite true!
That's why I've been so interested in other additional methods, that can come in to help this flimsy contraption. Something like detecting clipping or large distortion that appears in extreme cases, including when there is a short or an overly difficult load.
That detection is quite easy to do and foolproof, with a simple opamp, sensing the differences between the ltp's in+ and in-. Works perfectly.
Then using that info, something can be done to apply limiting right at the input of the amp, compressing the signal.
The VI protections have one major factor to act against, which make it more difficult to act strongly enough: they are inside the amp's feedback loop, so when they act, the amp fights against it and wants to keep the output up like the input is. So it's often a losing battle anyway. If there are enough output pairs, giving a lot of spare soa to play with, then it can be ok, but the prime goal of a more advanced VI limiter with dual slope, or more, in the first place, was to make use of more of the soa available. But that turns out hard to do in reality.
If the transistors could activate a little more like comparators, that would be much easier, with a sharper knee in their characteristics.
Do you mean caps in the VI limiter's circuit? In the sensing part?
I've played with that too, and in fact since there is already a delayed response and the action is weaker than it should be, that extra soa is already used by default.
What I've seen from the action of the limiters in case of shorts, is gross violations of soa, unless there is plenty of it in reserve. And that includes the reserve soa in short bursts.
So in fact, it turns out that decreasing the cap value on the limiter's bases, speeds things up a bit and it helps.
The limiter's transistors need to have a low Vcesat for the amp topos with fewer junctions to bias. It's less of an issue for the 3EF types, which need nearly 4V of total spread to bias.
For a topo with only a single junction on each side, there is less than 2V of total spread, so when the limiter activates, its Vcesat plus the diode in its collector have a hard time "squeezing" that well enough.
Ok, so this is the original leach amp, v4.5 with 2 pairs of MJ15003/4 and the drivers are the MJE15030/1 that came with the sim.
I ran the tests with an 8, 4 and 2 ohms resistive load, and then with a 1mohms short, and this at 1khz, and then at 20khz.
Obviously that amp with those outputs with only 2 pairs isn't for 2ohms load, and that also means not for a complex 4ohms, for which we allow dips down to 2ohms.
On 8ohms, it looks like plenty of reserve soa for the outputs, and for the drivers, there is a slight violation at 20khz.
On real signals, never full power at such frequencies, there is no chance it would happen.
However, we can see at lower load impedance, it can get a bit dangerous, and at 2ohms, the protections activate, at any frequency.
The ugly stuff happens on that 1ohm short, with a serious violation for the outputs at any frequency, no matter what the protections do, it's obviously nowhere near enough, and they do intrude on some legit signals.
I derated the soa (dashed line) for a 50C rise.
At 20khz on 8ohms, legit signal, no protection activation, but thd at almost 1%!!
I zipped up that sim, which hopefully should run out of the box.
I ran the tests with an 8, 4 and 2 ohms resistive load, and then with a 1mohms short, and this at 1khz, and then at 20khz.
Obviously that amp with those outputs with only 2 pairs isn't for 2ohms load, and that also means not for a complex 4ohms, for which we allow dips down to 2ohms.
On 8ohms, it looks like plenty of reserve soa for the outputs, and for the drivers, there is a slight violation at 20khz.
On real signals, never full power at such frequencies, there is no chance it would happen.
However, we can see at lower load impedance, it can get a bit dangerous, and at 2ohms, the protections activate, at any frequency.
The ugly stuff happens on that 1ohm short, with a serious violation for the outputs at any frequency, no matter what the protections do, it's obviously nowhere near enough, and they do intrude on some legit signals.
I derated the soa (dashed line) for a 50C rise.
At 20khz on 8ohms, legit signal, no protection activation, but thd at almost 1%!!
I zipped up that sim, which hopefully should run out of the box.
Attachments
-
Screen Shot 2017-03-22 at 6.13.26 PM.png -
Screen Shot 2017-03-22 at 5.57.00 PM.png -
Screen Shot 2017-03-22 at 6.00.06 PM.png -
Screen Shot 2017-03-22 at 6.01.02 PM.png -
Screen Shot 2017-03-22 at 5.55.12 PM.png -
Screen Shot 2017-03-22 at 5.48.22 PM.png -
Screen Shot 2017-03-22 at 5.47.25 PM.png -
Screen Shot 2017-03-22 at 5.46.09 PM.png -
Screen Shot 2017-03-22 at 5.55.35 PM.png -
Leach Amp v4.5 SOA.zip
Ok, I ran this one again. It hasn't changed since the earlier post.
It does run and the wave looks visually fine, until we look at different points inside the loop.
I left it as is, with 20khz freq, and I'm seeing thd at 0.157%. Are you getting the same thing?
And when looking at the signal on Q5's collector, there is that glitch on the negative side. That has to be a part of what contributes to the high thd.
FFT reveals how much hash there is, which has to account for that high thd.
Whatever that glitch is, generates harmonics.
The glitch is fairly small in comparison to the legit signal at the ltp output, but by the time it gets to the predrivers, it's the glitch that's far larger than the legit signal.
And yet, this is far better than what's going on in the superamp.
Attachments
You bring up the distortion often, I already mentioned that the Leach is not a high feedback
design and I don't believe that it ever simmed with low distortion. Show me a Leach sim
without significant mods that sims with low distortion. i'm not surprised because andy_c
has built them and mentioned years ago that they do not measure very well. Point me to
a thread with sims and/or measurments of a 4.5 Leach that shows otherwise.
Mine does not have those glitches, it has a very small amount in the predrivers around the
crossover point so I'm going to call that normal.
However, I fixed the duplicate models issue on previous sims but it is back on this one so
I can't be sure right now which models are being used.
My LTSpice install was done within the last 3 months without any mods.
I hate issues like this, find them very annoying.
design and I don't believe that it ever simmed with low distortion. Show me a Leach sim
without significant mods that sims with low distortion. i'm not surprised because andy_c
has built them and mentioned years ago that they do not measure very well. Point me to
a thread with sims and/or measurments of a 4.5 Leach that shows otherwise.
Mine does not have those glitches, it has a very small amount in the predrivers around the
crossover point so I'm going to call that normal.
However, I fixed the duplicate models issue on previous sims but it is back on this one so
I can't be sure right now which models are being used.
My LTSpice install was done within the last 3 months without any mods.
I hate issues like this, find them very annoying.
I'm now close to what you've got there, but what you have there is actually
reasonable for an output stage with crossover distortion. You're design has much
larger glitches with oscillations on them. I think that the BC550/560
are slow in the models.txt file which were being used and fixed the problem. Still,
and I'll repeat what you have there is reasonable with bad output device models.
When I replace your 2N3055/MJ2955 models with the MJ21193/4 it goes away.
But I'll comment on your comments here:
If you look closer at IcQ5 LTSpice expands the range of Ic so that the waveform
fills the graph. The actual variation around that glitch is not very large AND if
you look at it relative to the output waveform, you'll see that while it is at a peak
of Ic's current it is at the crossover point of the output and it is correcting for
crossover distortion in the output stage. It is phase shifted due to the Cdom cap.
The glitch in IeQ14 is right at the crossover point also and is due to the output models
I'm not going to spend any time chasing that down.
When I had models overlapping I didn't see that.
I'm going to guess that the BC550C in the models.txt file is very slow such that they
filter out those disturbances because that is the transistor I remember in the error log.
And probably Bob C's is more realistic.
I just confirmed that 2n3904/6 do it and several others in place of the input devices.
Still this sim is much more reasonable, I expect such behavior in the crossover region
with low bias or bad devices.
Attached is the sim with the aliased BC550/60C model issues fixed AND with Bob C's
MJ21193/4 outputs ALL MODELS ARE BOB C's. NO GLITCHES. I have probably
repeated this 4 times in this thread yet you seem to have a fascination with bad models.
So continue as you've been doing and enjoy your frustration. THD is only very slightly
better at .15%.
reasonable for an output stage with crossover distortion. You're design has much
larger glitches with oscillations on them. I think that the BC550/560
are slow in the models.txt file which were being used and fixed the problem. Still,
and I'll repeat what you have there is reasonable with bad output device models.
When I replace your 2N3055/MJ2955 models with the MJ21193/4 it goes away.
But I'll comment on your comments here:
If you look closer at IcQ5 LTSpice expands the range of Ic so that the waveform
fills the graph. The actual variation around that glitch is not very large AND if
you look at it relative to the output waveform, you'll see that while it is at a peak
of Ic's current it is at the crossover point of the output and it is correcting for
crossover distortion in the output stage. It is phase shifted due to the Cdom cap.
The glitch in IeQ14 is right at the crossover point also and is due to the output models
I'm not going to spend any time chasing that down.
When I had models overlapping I didn't see that.
I'm going to guess that the BC550C in the models.txt file is very slow such that they
filter out those disturbances because that is the transistor I remember in the error log.
And probably Bob C's is more realistic.
I just confirmed that 2n3904/6 do it and several others in place of the input devices.
Still this sim is much more reasonable, I expect such behavior in the crossover region
with low bias or bad devices.
Attached is the sim with the aliased BC550/60C model issues fixed AND with Bob C's
MJ21193/4 outputs ALL MODELS ARE BOB C's. NO GLITCHES. I have probably
repeated this 4 times in this thread yet you seem to have a fascination with bad models.
So continue as you've been doing and enjoy your frustration. THD is only very slightly
better at .15%.
Attachments
Last edited:
I don't know of any, and I suspect there aren't any unless some serious alterations have been made.
I think it's mostly because of the glitches I've been having in those sims, which obviously inflate the thd result, but I think that if those glitches are suppressed, then the thd, although still not stellar in performance, would likely be much lower than what I've seen.
Since I've had thd at levels of about 1% and up, it's rather high, even for a leach amp.
That must be with changes in models that you've made right?
Which sim was that? If it's the same as in post #188, then nothing has changed, and most of the models are Bob's, except for the 3055s and the MJE drivers, which were from leach's sim.
I swapped the MJEs with Bob's and although they're different and that changes the operating point, it doesn't change anything about glitches, so those models don't seem to be the cause, and they don't seem bad.
I can't say the same, as there hasn't been any updates for me for quite some time and it looks like there won't be any any time soon as well.
Since I've been running ltspice for quite some time, with no updates and some models were needed so many times long ago that weren't in the main library, to avoid all that constant shuffling of model files, at the time I had added a few in the main library. But I have since been disabling them. Because of the needs for sharing, I had to start again using independent model files.
It is. If there are no more ltspice updates that I can use, only the disabling of models in its library can help with the mess.
Those are Bob's, and the only possible change that eventually could've been made is the BF (beta) to make them more like the real thing and less dissimilar.
I had tried that, the 3055s swapped with Bob's 21193/4 and the MJEs for Bob's 32/33, and that didn't change things to get rid of the glitches. So I'm wondering if there could be more than just the models going on...
I haven't used any other 550/60 models than Bob's.
Excellent. But no sim was attached.
And I was using Bob's models, except for the outputs and drivers.
No specific fascination, just the goal being a 3055 based superamp and there aren't any such models in Bob's lib, I use what I have handy.
Well that's much better than some 1% or more.
Now if we can get this sim working with protections, we can start thinking pcb afterwards.
We are going in circles here because I say USE ALL BOB C's models and you respond with:
"And I was using Bob's models, except for the outputs and drivers."
EXCEPT FOR THE OUTPUT AND DRIVERS is your key statement.
This is probably the 5th time I've said try Bob's models and what do you know, they
fix the glitching, conclusion there is a problem with the models that you insist on
using. You say that the BC550/560 are Bob's but with a minor tweak, hmmm I'm able
to sim without glitches and you are not, hmmm, let's think about that.
You seem to want to follow your own course, that is certainly an option for you but I
can't help you if that is the case.
I've been through this type of frustration using defective models, I explained that and
stated how Bob's models work, you don't seem to be a believer so there is no point in
going further.
You need to either read Bob's chapter on fixing SPICE models and do the entire procedure
on your defective models or just use the MJLs knowing that you have to build a prototype
to confirm the performance with whatever older parts you want to use. I am of the opinion
that you have to build a prototype no matter what.
You have the SIMP-DIST version file, if you change the drivers, use all BOB C's models
out of his models file it should work the same as mine without glitches or oscillations.
You have to use BOB's MJL outputs.
I would NOT go to PCB you should build a prototype and prove out a real world build.
Stability analysis is needed etc. I'm simming SIMP-DIST not your beta multiplier, cascode
version and offering SIMP-DIST as an example of a design that does not oscillate or glitch.
I don't think that your version works without glitches or potential instability.
One other thing on my mind is that the TIAN probe might not be in the correct spot, or
possibly due to nested feedback loops it is not giving correct results. I have to think
about this more.
Attached is the SIMP-DIST zip file with all the files to run what does not glitch or oscillate.
I call it DIST because it is my mods for lower distortion, which you complained about,
so it is a Leach 4.5 with my minimal mods for lower distortion. I would not build this with
BD139/140 but they are close enough to med Cob modern parts for simulation. There
are modern subs that I'd use some with much lower capacitance. This still needs detailed
stability analysis and fine tuning. I'm posting this more for other people who might want
to build something with 2N3055s or a Leach 4.5 with somewhat lower distortion, or just
try out a simulation.
"And I was using Bob's models, except for the outputs and drivers."
EXCEPT FOR THE OUTPUT AND DRIVERS is your key statement.
This is probably the 5th time I've said try Bob's models and what do you know, they
fix the glitching, conclusion there is a problem with the models that you insist on
using. You say that the BC550/560 are Bob's but with a minor tweak, hmmm I'm able
to sim without glitches and you are not, hmmm, let's think about that.
You seem to want to follow your own course, that is certainly an option for you but I
can't help you if that is the case.
I've been through this type of frustration using defective models, I explained that and
stated how Bob's models work, you don't seem to be a believer so there is no point in
going further.
You need to either read Bob's chapter on fixing SPICE models and do the entire procedure
on your defective models or just use the MJLs knowing that you have to build a prototype
to confirm the performance with whatever older parts you want to use. I am of the opinion
that you have to build a prototype no matter what.
You have the SIMP-DIST version file, if you change the drivers, use all BOB C's models
out of his models file it should work the same as mine without glitches or oscillations.
You have to use BOB's MJL outputs.
I would NOT go to PCB you should build a prototype and prove out a real world build.
Stability analysis is needed etc. I'm simming SIMP-DIST not your beta multiplier, cascode
version and offering SIMP-DIST as an example of a design that does not oscillate or glitch.
I don't think that your version works without glitches or potential instability.
One other thing on my mind is that the TIAN probe might not be in the correct spot, or
possibly due to nested feedback loops it is not giving correct results. I have to think
about this more.
Attached is the SIMP-DIST zip file with all the files to run what does not glitch or oscillate.
I call it DIST because it is my mods for lower distortion, which you complained about,
so it is a Leach 4.5 with my minimal mods for lower distortion. I would not build this with
BD139/140 but they are close enough to med Cob modern parts for simulation. There
are modern subs that I'd use some with much lower capacitance. This still needs detailed
stability analysis and fine tuning. I'm posting this more for other people who might want
to build something with 2N3055s or a Leach 4.5 with somewhat lower distortion, or just
try out a simulation.
That latest .zip has a truncated models.txt file, you can probably run without it even
since I'm using the MJLs as outputs.
There is so much back and forth with me saying use Bob's models and you saying I did
EXCEPT, that I don't think I can help you. All the time is wasted on this.
I have my own ideas of how to go from sim to real hardware and perhaps you have
something different and I really don't have time to explain my methods. First, I get the
sim to run without issues, last thing is to worry about distortion.
since I'm using the MJLs as outputs.
There is so much back and forth with me saying use Bob's models and you saying I did
EXCEPT, that I don't think I can help you. All the time is wasted on this.
I have my own ideas of how to go from sim to real hardware and perhaps you have
something different and I really don't have time to explain my methods. First, I get the
sim to run without issues, last thing is to worry about distortion.
I've already commented on SOA limiters and how I don't like them, you were rather dismissive of that. Yet you can't seem to get them to work. They are flawed IMO since
transistors do not make good comparators. Perhaps with enough margin they could be
made good enough but I'm not going to spend a lot of time on them.
Fuses were too crude for you, go ahead and make the SOA limiter work.
I also mentioned a sense transistor on the emitter resistor that puts the amp into protect
mode, which you also dismissed. I would set this to trip at less than 1 or .5 ohms to lock
up and protect the amp from a short. Yes everything stops until the short is fixed. If a
person puts so many pairs of speakers in parallel to present 1 ohm, then it should lock
up since it is not designed for that. If the amp is designed for 1 ohm then you obviously
set the trip point lower. I'd use this in conjunction with SOA or fuses.
I've put a lot of thought into this and drawn from my own experience.
transistors do not make good comparators. Perhaps with enough margin they could be
made good enough but I'm not going to spend a lot of time on them.
Fuses were too crude for you, go ahead and make the SOA limiter work.
I also mentioned a sense transistor on the emitter resistor that puts the amp into protect
mode, which you also dismissed. I would set this to trip at less than 1 or .5 ohms to lock
up and protect the amp from a short. Yes everything stops until the short is fixed. If a
person puts so many pairs of speakers in parallel to present 1 ohm, then it should lock
up since it is not designed for that. If the amp is designed for 1 ohm then you obviously
set the trip point lower. I'd use this in conjunction with SOA or fuses.
I've put a lot of thought into this and drawn from my own experience.
Attached is the LTSpice Sim of the new series topology that I came up with, no glitches
or oscillations. This is not for commercial use but is for builders who for some reason
need to use older lower voltage output devices. I'm calling it "BASEL SUPER SERIES
POWER AMP DERIVED FROM LEACH SUPERAMP 3-23-2017" because it is different
enough from any of the other amps out there with series output stages.
Sims are with all Bob C. models and MJL21193/4 outputs which are good enough to
prove out the design - a prototype should be built if anyone takes this further. This is a
start, stability analysis Pdiss, and replacements for the BD139/140 need to be considered.
Better feedback such as TMC, MIC or others should be considered.
Use the attached .asc files with the support files in the previous .zip.
It sims with .16% THD20 slightly below full clipping.
or oscillations. This is not for commercial use but is for builders who for some reason
need to use older lower voltage output devices. I'm calling it "BASEL SUPER SERIES
POWER AMP DERIVED FROM LEACH SUPERAMP 3-23-2017" because it is different
enough from any of the other amps out there with series output stages.
Sims are with all Bob C. models and MJL21193/4 outputs which are good enough to
prove out the design - a prototype should be built if anyone takes this further. This is a
start, stability analysis Pdiss, and replacements for the BD139/140 need to be considered.
Better feedback such as TMC, MIC or others should be considered.
Use the attached .asc files with the support files in the previous .zip.
It sims with .16% THD20 slightly below full clipping.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.