Guess that is the way to go. Use Excel to organize all my acquired models and let the standard library do it's thing. Just not as convenient.
Sharing. I hear you. The problem is most of them I have stripped the copyright comments ( when LTSpice cleans them up) so it is inappropriate to put them out there. Most came from OnSemi, Central, Fairchild THAT, Exicon etc, so they are available. I have not measured any myself. I did generate more complete values for one series of 1N series zieners based on parameters from two sites and some interpolation. A few diodes seemed to work too well and sure enough, the model had no capacitance. At least that is one you can gustimate from the data sheet. ( default is zero) For a few, I am adding the display text for voltage and current.
I guess most vendors expect us to just have a directory of their "contents of one" as we should know what we are looking for.
Sharing. I hear you. The problem is most of them I have stripped the copyright comments ( when LTSpice cleans them up) so it is inappropriate to put them out there. Most came from OnSemi, Central, Fairchild THAT, Exicon etc, so they are available. I have not measured any myself. I did generate more complete values for one series of 1N series zieners based on parameters from two sites and some interpolation. A few diodes seemed to work too well and sure enough, the model had no capacitance. At least that is one you can gustimate from the data sheet. ( default is zero) For a few, I am adding the display text for voltage and current.
I guess most vendors expect us to just have a directory of their "contents of one" as we should know what we are looking for.
I suppose this would be an appropriate forum for my spice question.
I was trying to simulate a grounded bridge amp, using a single power supply without a center tap, and I can't figure out how to make this work properly.
Using a single voltage source, set up as DC to provide the rails and have a separate ground reference seems impossible, or I just don't know how it can be done.
I tried that and it (ltspice) wants to put the minus side of the power source as the ground reference level, so this can't work as the ground must float freely between the rails.
How can this be done? If at all.
I was trying to simulate a grounded bridge amp, using a single power supply without a center tap, and I can't figure out how to make this work properly.
Using a single voltage source, set up as DC to provide the rails and have a separate ground reference seems impossible, or I just don't know how it can be done.
I tried that and it (ltspice) wants to put the minus side of the power source as the ground reference level, so this can't work as the ground must float freely between the rails.
How can this be done? If at all.
Your amp needs a reference between the rails for audio ground right? For a single-rail supply this would always have to be done with a virtual ground, which can be as simple as two 1k resistors (resistive divider) across the rails where their centerpoint is the ground. Put the LTSpice ground here and all will be well.
I don't know how bridge amps usually do it, they may reference the negative rail for input ground if that is the intended configuration for the chip/circuit. You should look at where signal ground is in the app note/datasheet if there is one.
I don't know how bridge amps usually do it, they may reference the negative rail for input ground if that is the intended configuration for the chip/circuit. You should look at where signal ground is in the app note/datasheet if there is one.
Yes, and it's provided by the low side of the grounded bridge, not by a center tap from a psu.
Can't do that! The ground is the output of the low side in the bridge and I do need a center tap for some references, but those aren't ground. And the psu in a grounded bridge is without a center tap. The low side grabs on to the ground, makes it the reference and makes the rails fly around it.
My spice problem is that using a single power source doesn't work because spice makes the negative end of that source as the ground reference, and that doesn't work because it's actually supposed to be left flying as the negative rail. The ground can't come from a power source, it's set by the low side amp.
There isn't any, it's a discrete amp, all bjts. It's a bridge, but not the classic one, it's a grounded one, with one side's output bound to and setting the ground and thus the reference.
So spice is giving me an issue with that, if I want to use only one power source as it's supposed to be, then it won't let me.
That is fine, and having a single power source should work, as long as it's not trying to force the ground on its negative end, and thus shorting that rail to the ground, which is set by the amp's low side output.
It does work if I put 2 power sources in series, but then there is a center tap.
For bridged amp the output is basically centered by divider regardless of whether you explicitly add one - the output has to be between the rails to get full output voltage. So by adding a resistor divider (AKA virtual ground) as a reference you're not ruining anything. It's probable this divider is already on the schematic if you look for it (although it can be done with zeners or transistors instead of resistors).
I'm referring to the grounded bridge as designed by crown a long time ago.
To make sure we're on the same wavelength, here's a link to their paper about it:
http://www.crownaudio.com/media/pdf/amps/grbgpapr.pdf
You can see on page 6 on that document how the psu is very simple, no center tap, totally floating, and it's the low side of the bridge that sets the ground reference and it forces the rails to float around that ground.
In the amps that I'm trying to bring together as such a bridge, it takes 2 power sources to make it work, as using one is giving the issue I describe. I can use that center tap as a reference, that isn't the ground but rather the middle point of the psu, or I can make up a divider for that, but that point is not the ground.
If using a single power source as it's meant to be, it won't work because of the negative end of that source being set as the reference to ground, which shorts half the amp, because the negative rail should swing as the positive does and the reference to ground is not set by the psu but by one side of the amp.
That's what I've been trying to solve, unsuccessfully so far. And I'm wondering if there really is a solution to this.
As long as all the current flows through the load, not ground, then you only need a virtual ground as reference for the feedback network; this is unique to a bridge amp. It's hard to say how this could be done without seeing the schematic, though it seems you have good reasons not to share it.
The grounded bridge is just a bridge amplifier. It's just configured in a way such that one side of the speaker is referenced to signal ground. Its biggest advantage (over a "normal" bridged amplifier) is probably simpler biasing of the voltage gain stages and the feedback network.
Please note that the description of Class-H in that document you linked is wrong. (And that what they call Class-H, Europeans and Japanese tend to call Class-G). In their figure 9, transistor Q1 does not enter saturation before Q2 turns on (well, not if the amplifier is designed properly anyway). There is indeed additional signal distortion as they show in their figure 10, but this is caused by the crossover from diode to Q2, not by Q1 entering/leaving saturation.
Going back to your issue, it would be much easier to help you if you could post a schematic.
Since this thread is aimed at spice and not really at any specific circuits, my question was about that spice issue and I didn't think I'd need to post a schematic. But there is nothing secret about that simulation, it's the 2N3055 grounded bridge that we're trying to build in an other thread.
Actually I am trying this on more than one topology, still a grounded bridge, but different amps.
But here is the example, I'm attaching one of the sims (with the models file used) that I'm trying to get going. This one is the main one we've been working on since the beginning of this project, based on the elektor amp from way back. And I am also working on john ellis' very neat bridge as well.
With this particular amp, the bridge is using a divider network exactly as crown uses in their amps (see the dotted box on the low side). And there is no need for any other reference from the rails. The other reference is the ground and that is the low side amp that creates it.
What I had in mind for simulation was to add the whole psu, with a rectifier bridge and 3 big filter caps between the rails. I even though about simulating with a transformer, but that requires a model that I don't even know where to get. So I figured if I can use a sine wave voltage source and rectify that, then I would get my rails. But that didn't work. So I backed up to using only a single DC voltage source that would be the rails, but then the issue of the reference came up.
The power supply is also much simpler, without a center tap, and doesn't make any reference to ground.
This provides one further advantage: the grounded bridge can be further bridged. As 2 grounded bridges can then be bridged (a bridged bridge).
Well, they were just trying to compare with other topos. I guess it's mostly a matter of name calling. To me the only difference between the class H and G is the way the rails are switched.
I just did (one of them). I'm using ltspice.
A "normal" bridge amplifier also does not require a centre tap for the PSU transformer secondary. I.e. it needs only a single power supply instead of a dual centred around 0 V.
How? Can you draw a simplified schematic? What would be the purpose of doing this?
Whether they call it Class-G or Class-H, their description of how the circuit operates is incorrect.
Indeed. If you look the inner device (Q1 in Crown's paper), as the output signal increases from 0 V, Q1 Vce decreases. When it reaches about 5 V, Q2 will start to come on. In the Euro/Japanese convention, Class-G means that Q2 is controlled such that Q1 Vce now stays at a constant 5 V until the amplifier clips, and Class-H means that Q2 is turned hard-on (saturated), immediately pulling Q1 collector to the higher rail. In both cases, Q1 does not saturate until the amplifier clips (output signal gets to within a few volts of the highest power supply rail). Class-H tends to be fractionally more efficient but there are bigger issues at the transition point in terms of the switching from one rail to another feeding through to the output.
In most American publications, these definitions of Class-G and Class-H are transposed.
You may be mixing up the conventional full bridge and the grounded one.
A regular full bridge doesn't make its rails "float around" the ground, but a grounded one does.
I don't have anything handy at the moment, I'll have to look this up...
I guess you're not aware of the difference and what a grounded bridge really is.
In a grounded bridge, two sides (amps) are bridged together, but with on side (low side) having its output tied to ground, and it's the one grabbing on to that ground and forcing the rails to fly around it.
A grounded bridge is made of 2 amps, with one a little simpler than the other (the low side), and it's made to be working as a single amp, always bridged, but since it refers to ground on one side, the other being the output, then it can again be bridged (a bridged bridge), so in fact, there are 4 amps working together, on a single load.
And the purpose, is simple, it allows making powerful amps that don't require transistors with as high vce0. The power output quadruples when bridging, and then quadruples again when bridging the bridge.
This is what crown has be doing for a long time.
Maybe, but the class H and G concepts are not what makes a grounded bridge
. It's totally different. It's true that one advantage is being able to use far higher rail voltages with lower vce0 transistors, with class H, G and with a grounded bridge, but the grounded bridge topo is ordinary class AB.
We're working on making a grounded bridge with 3055s, with rails at 35V max. So this means we might get close to 200W on 8ohms, of the psu doesn't sag too much under load. This is a fair amount of power, considering that we're only using 3055s and only 35V rails and nothing like class H or G. Then if done right, that bridge can again be bridged and could get probably some 700W of power, depending on the psu losses.
The 35V rails on a single amp gives something like 20Vrms on the load (50W/8ohms). The bridged one will double that (40Vrms, 200W/8ohms), but that 8ohms load on the bridge is like 4ohms on each side, so there would be more rail sagging. Then bridging the bridge doubles that voltage swing again, but this time the amps see a 2ohms load each, which would cause even more sagging, however still a fair amount of power.
The psu can be made stiff enough, if it's planned to provide that much power, too much sagging can be limited.
I test for different things, including square wave behavior and I don't want to make a bunch of separate sim files, so I put it in there and when not in use, it's better shorted than hanging as a floating node. Since it's shorted and not interfering with anything else when not hooked up, it doesn't bother other simulations. When I want to simulate with it, I hook it up properly, short something else where needed and there you go.
are you also using coconuts and bamboo to make the rest of the amp?
MJL21193/4 are real improvements in all specs for slow rugged audio BJT and have better thermal interface
and the mode today is probably to use much faster RET output Q paralleled for the needed SOA
using 2N3055 for audio today is an exercise in Creative Anachronism
No, I'm not. A conventional bridge does not require a dual power supply.
Yes, I understand.
What I don't get is how you can bridge a bridge. Presumably the configuration ends up effectively placing the two independent and floating supplies in series, doubling the available supply voltage.
Yes I know. I understand that this was/is a tangent and the Crown paper you linked only mentions Class-H/G as a comparison. I just felt that it was worth mentioning, for anyone who might read that document, that its description of operation of Class-H is wrong.
Indeed. The 2N3055 is spectacularly horrid. The MJL21193/4 is still not great compared to the ubiquitous MJL1302/3281.
With the low Vceo requirement, the amplifier could possibly use the 2SA2222SG/2SC6144SG. These are very fast and have low enough parasitic capacitance that even with the number you'd need in parallel to give the required power handling, you'd still end up with a faster output stage than using larger transistors such as the 1302/3281. This in turn would allow for a higher unity-gain bandwidth and a higher feedback factor (lower distortion) in the audio band.
Ah ah! funny!
I'll design a nice metal case, using cad.
True. But the point and goal there is to use 3055s, and we can get the best out of those oldies and have diy fun in the process. Those are still around, I have lots of them in stock, aching to go to work, and there are places on this planet where such things are common and the better ones not so much.
Could be, and that's what we're doing!