you are correct, i was referring to the active CCS proposed by Lazy Cat...
those resistor tails are good enough for me, i wouldn't change them....
I think the leach amp is a totally proven, tested, in large numbers and found to be quite high quality.
I am of course open to new things, but before going for some other design, I would want it to be tested properly, a prototype made, many comparisons by as many people as possible to make sure not only that it does work right, always, and that there are actual improvements that were made and that brought something more.
We can't go wrong with the leach as it is right now.
We can make a few minor alterations without leaving the realm of the philosophy behind it.
The changes that revert the low tim aspect to some other "type" of concept makes that amp some other amp and it's no longer leach's amp.
One thing that was done purposely in that amp to aim for low tim, was to reduce gain at each stage and apply a moderate feedback overall. Each stage thus has local feedback that's calculated specifically to reduce that gain, with other few benefits in the process, and leach has even added a feedback from an intermediate stage. Everything he has done was well thought out and very carefully considered and tested.
I don't think anyone out there who has built the leach amp has ever found its sound quality wanting.
Of course nothing is ever perfect and I'm sure we can probably improve on things. So if such improvements do indeed bring something, then why not. But before simply switching to it, I'm sure most will agree that proper testing must be done and proper comparisons will tell.
I am of course open to new things, but before going for some other design, I would want it to be tested properly, a prototype made, many comparisons by as many people as possible to make sure not only that it does work right, always, and that there are actual improvements that were made and that brought something more.
We can't go wrong with the leach as it is right now.
We can make a few minor alterations without leaving the realm of the philosophy behind it.
The changes that revert the low tim aspect to some other "type" of concept makes that amp some other amp and it's no longer leach's amp.
One thing that was done purposely in that amp to aim for low tim, was to reduce gain at each stage and apply a moderate feedback overall. Each stage thus has local feedback that's calculated specifically to reduce that gain, with other few benefits in the process, and leach has even added a feedback from an intermediate stage. Everything he has done was well thought out and very carefully considered and tested.
I don't think anyone out there who has built the leach amp has ever found its sound quality wanting.
Of course nothing is ever perfect and I'm sure we can probably improve on things. So if such improvements do indeed bring something, then why not. But before simply switching to it, I'm sure most will agree that proper testing must be done and proper comparisons will tell.
Hi Spookydd you have a great project going. Remember you cannot please everyone, nor can you factor in everything in a single design. The Leach amp has a long history, many engineers were taught using that amplifier. Do try it out. You have come pretty far. There will be time more for other designs including Lazy Cats and SYMEF.
kind regards,
Harrison.
kind regards,
Harrison.
That's what I'm thinking, and I'm trying not to deviate too much from the intended design. My small alterations are really minor and don't detract from the soul behind it, I hope.
This may please a fair number of amateurs of that amp, so why try to go into uncharted territory right now and lose interest?
If it wasn't as good as it is, I doubt it will still be so popular now, after all these years. I'm sure this one will still be built by many in many years from now, and many new designs from this current period will have long been forgotten along the way.
Unless I can find something truly better, not just because someone says it, but because it was tried and proven by many, I don't see any others that are up to the standard right now. And as I mentioned before, I am not interested in mosfet based ones, and although I think some "classes" are rather interesting, such as G/H, A and for some, even D, I'm sticking with AB for now.
Absolutely! I am considering those possible "updates", but that would be for a future version, and it needs testing and prototype.
I think such changes can rather easily be brought in the design without having to radically change everything. I like this current design. I gave it a lot of thoughts.
It's not easy to find all the right compromises. I'm trying to make it as elegant as possible, as easy as possible to repair if need be. And hopefully making it good enough so repairs would be as unlikely needed as they possibly can be.
I think the extensive testing and research done by our friend Bonsai on the mosfet based SSRs is excellent and well documented. He's done a great job and I am convinced now to make use of this. So I am changing all the relays and fuses for SSRs (although not crowbars), because I do think it's superior and far better and efficient protection. With a good protection and things like input muting and perhaps input limiter or even compressor to prevent clipping, this amp can be as indestructible as for example the crown amps.
btw: The is one concept that I find quite interesting (sch attached).
Although this is more of a testing circuit to simulate and perhaps it could be built as a prototype to find out how it works in reality. This has certain aspects that I find very interesting. The SQ remains to be seen and tested.
Perhaps this could be done all discrete and remove that opamp, and eventually a more conventional topo (such as leach's) could be adapted to this scheme.
I'm unfortunately not geared for such prototyping and research, but perhaps there are other enthusiasts that could tackle this.
Interesting this, is that like in a grounded bridge, the output is grounded!
One very interesting feature, is the fact that all outputs have their collectors, not only together, but also grounded. This is great for heatsinking, the cases, no need for insulation, no danger of touching high voltages on the transistors, etc...
This is nice, because TO3s can be used directly without shims (no mica or other), so we can have a better heat exchange situation, possibly reducing the required heatsink size.
The wiring can also get interesting, with the amp's output becoming the central star ground and the speaker's other end goes to the center tap on the psu...
Many cool things to learn from this. (thanks QSC)
Although this is more of a testing circuit to simulate and perhaps it could be built as a prototype to find out how it works in reality. This has certain aspects that I find very interesting. The SQ remains to be seen and tested.
Perhaps this could be done all discrete and remove that opamp, and eventually a more conventional topo (such as leach's) could be adapted to this scheme.
I'm unfortunately not geared for such prototyping and research, but perhaps there are other enthusiasts that could tackle this.
Interesting this, is that like in a grounded bridge, the output is grounded!
One very interesting feature, is the fact that all outputs have their collectors, not only together, but also grounded. This is great for heatsinking, the cases, no need for insulation, no danger of touching high voltages on the transistors, etc...
This is nice, because TO3s can be used directly without shims (no mica or other), so we can have a better heat exchange situation, possibly reducing the required heatsink size.
The wiring can also get interesting, with the amp's output becoming the central star ground and the speaker's other end goes to the center tap on the psu...
Many cool things to learn from this. (thanks QSC)
I haven't been working much on this project lately because I've been working on a grounded bridge with 2N3055.
I am however giving it some thoughts and I have decided to do away with the VI limiter type of protection and replace this with a more comprehensive protection scheme, using solid state switches as eminently described by our friend Bonsai:
Solid State Loudspeaker Relays for Audio Amplifiers
His research work on this subject has convinced me that this is a much better way to protect amps than any other methods.
Thus I will not use any output relays and VI limiters, in favor of solid state switches on the output and on both rails as well. Plus I want to add an input muting, with the proper logic to have proper timing for switching those in the right order.
I am however giving it some thoughts and I have decided to do away with the VI limiter type of protection and replace this with a more comprehensive protection scheme, using solid state switches as eminently described by our friend Bonsai:
Solid State Loudspeaker Relays for Audio Amplifiers
His research work on this subject has convinced me that this is a much better way to protect amps than any other methods.
Thus I will not use any output relays and VI limiters, in favor of solid state switches on the output and on both rails as well. Plus I want to add an input muting, with the proper logic to have proper timing for switching those in the right order.
The link in my previous post is Bonsai's excellent work on SSR, and that's what I'm going to implement, on output and rails.
For the amp's soa protection, I'm thinking about other ways not using vi limiters. For one thing, I want to oversize the output stage, so it can withstand a serious amount of abuse to begin with. Then I want to figure out a good way to limit the input signal to prevent the amp from ever clipping, but gain control, perhaps using a good opamp before the amp's input and sense some level, either on the output directly or some other way, to control that gain so the signal sent to the amp's input can never allow it to clip.
This input gain control against clipping can also double up as a mute, by bringing the gain to 0.
The only thing left to protect the amp from overload is the over-current when using very low load below 4ohms.
I don't want crowbars and I prefer not having any complete stop unless there really was a serious failure. A simple overload is not a reason for an immediate shutdown. The show must go on, unless the amp has blown. Then obviously if the amp is blown, the speakers must be protected and of course the show can't go on that way.
The amp must have protections, especially against its own failures, so the speakers don't get hurt. But I want to look for ways other than the vi limiter.
Plus I want a comprehensive turn-on/shut-down and protection, with the proper timing so input and output muting is done in the right order wether coming up or going down, in case of normal use or failure.
All this could be done with a micro-controller (arduino for example), as some have done, very neatly too, but I don't want to go that route (yet) and prefer plain discrete logic to handle that.
I want to make a nice amp, with all the trimmings, but not make it a Rube Goldberg contraption (usine a gaz).
One other thing that I wanted to mention is that since I have been getting into ltspice simulation for many weeks now and I'm trying many designs that way: I have tried a simulation of the leach amp and I was expecting a better performance in such sims.
I may not be using the best models and may obviously not be doing everything right, but since I have simulated many others before that one and its performances in sims are much much lower, either I am missing something, or doing it wrong.
At first, the thd was close to 1%. Granted, that was at 20khz, but on 8ohms only and at nominal power level. I have been getting 10s of ppm on other designs, even using 3055s (the grounded bridge project), so how come the leach amp isn't performing better in sims???
There is one thing that did improve the thd performance in the leach sim, and that was to make use of other transistors, such as BC550/60C for the diff amps and in the vas, and the mjl4281/4302 instead of mj15003/4 on the outputs. I switched the bc546b for a bd139 on the bias spreader and mj15032/3 for drivers.
Using the MJE340/50 for the vas and pre-drivers didn't work very well, so I used the ksa1381/ksc3503.
Still, better transistors helped, but the thd isn't that great. How come?
I may not be using the best models and may obviously not be doing everything right, but since I have simulated many others before that one and its performances in sims are much much lower, either I am missing something, or doing it wrong.
At first, the thd was close to 1%. Granted, that was at 20khz, but on 8ohms only and at nominal power level. I have been getting 10s of ppm on other designs, even using 3055s (the grounded bridge project), so how come the leach amp isn't performing better in sims???
There is one thing that did improve the thd performance in the leach sim, and that was to make use of other transistors, such as BC550/60C for the diff amps and in the vas, and the mjl4281/4302 instead of mj15003/4 on the outputs. I switched the bc546b for a bd139 on the bias spreader and mj15032/3 for drivers.
Using the MJE340/50 for the vas and pre-drivers didn't work very well, so I used the ksa1381/ksc3503.
Still, better transistors helped, but the thd isn't that great. How come?
The different transistors in an amplifier do different jobs.
The designer should select suitable transistors to suit each duty.
Leach lists the transistors, both in the original papers and in updates using more modern devices.
Selecting models for a simulator is another quite different problem.
The designer should select suitable transistors to suit each duty.
Leach lists the transistors, both in the original papers and in updates using more modern devices.
Selecting models for a simulator is another quite different problem.
SSR's in the rails? Interesting, but you could just use a pass transistor (since it's not an AC signal) like a regulator and have them shut down on fault. The catch I see there is if you shut down the rails you have to go all the way through the startup process to proved enough time for the front end to stabilize. Unless of course you are just using that instead of a crowbar.
VI limiters, especially triple slope ones, allow you to make use of most of the output's SOA. Their purpose is the same as "protect the amp from overload is the over-current when using very low load below 4ohms". Poorly implemented they can intrude on the sound, but done properly with an appropriately sized output stage you won't hear them most of the time.
For a commercial amp, VI limiting makes a lot of sense. For DIY, we can live without it if you think it interferes with the sound and understand/accept the risk. (Yes, I know some will vehemently disagree) I have built a number of amps some with VI protection and some without. I've never included a speaker relay. My amps (Leach, Honey Badger, A75) don't have turn on thumps and I've never blown a speaker. (well once when I dropped a screwdriver into the amp, but never once finally adjusted and the top went on.) I've never blown an amp (even without VI protection), but I am very careful about avoiding shorted speaker cables.
As for timing the sequence, while nothing says you can't mute the input on startup, as long as the speaker is disconnected, who cares? Turn on, stabilize a second or so, connect the speaker. Look at the system power up sequence if that still gives you thumps. Or set the speaker on delay long enough that all the components in the chain have had time to stabilize.
I am in the Keep It Simple Stupid camp. It seems to me that you are looking to do partially reinvent the wheel and over-thinking the problems. Nothing wrong with a belt and braces approach, but do we really need it in DIY?
It's your amp, your choices. You know where I stand.
The real purpose of my last post was just to make sure that you understood that your speaker SSR was not going to protect your amp from shorts.
VI limiters, especially triple slope ones, allow you to make use of most of the output's SOA. Their purpose is the same as "protect the amp from overload is the over-current when using very low load below 4ohms". Poorly implemented they can intrude on the sound, but done properly with an appropriately sized output stage you won't hear them most of the time.
For a commercial amp, VI limiting makes a lot of sense. For DIY, we can live without it if you think it interferes with the sound and understand/accept the risk. (Yes, I know some will vehemently disagree) I have built a number of amps some with VI protection and some without. I've never included a speaker relay. My amps (Leach, Honey Badger, A75) don't have turn on thumps and I've never blown a speaker. (well once when I dropped a screwdriver into the amp, but never once finally adjusted and the top went on.) I've never blown an amp (even without VI protection), but I am very careful about avoiding shorted speaker cables.
As for timing the sequence, while nothing says you can't mute the input on startup, as long as the speaker is disconnected, who cares? Turn on, stabilize a second or so, connect the speaker. Look at the system power up sequence if that still gives you thumps. Or set the speaker on delay long enough that all the components in the chain have had time to stabilize.
I am in the Keep It Simple Stupid camp. It seems to me that you are looking to do partially reinvent the wheel and over-thinking the problems. Nothing wrong with a belt and braces approach, but do we really need it in DIY?
It's your amp, your choices. You know where I stand.
The real purpose of my last post was just to make sure that you understood that your speaker SSR was not going to protect your amp from shorts.
Re: Simulations
Look at the power in your VAS and predrivers - oops. The BCs will work in sims but would blow in real life. You get lower distortion with them because they are higher gain devices than those suitable for VAS/predriver. Try the toshibas used in the Honey Badger, they have a bit more gain. Also try MJE15034/5 for a bit more hfe than the 15032/3.
Look at the power in your VAS and predrivers - oops. The BCs will work in sims but would blow in real life. You get lower distortion with them because they are higher gain devices than those suitable for VAS/predriver. Try the toshibas used in the Honey Badger, they have a bit more gain. Also try MJE15034/5 for a bit more hfe than the 15032/3.
The mosfets are easier to handle, with optocouplers, so the command circuits can be fairly easy, plus they have low rdson, which makes them better suited as switches than bjts, especially when it comes to dissipation (I don't see adding an extra heatsink for them as a good thing). They can handle a huge current and even voltage, compared to bjts, so I think they're a clear choice for that purpose. And since the rails are DC, then the switches can be simpler than what's used on the amp's output.
I don't want crowbars, that's too radical and they could turn out into a catastrophe if the amp doesn't have proper soa protection. I'd rather protect in other ways.
They go with an over-sized output stage, so they can be designed to never ever come into play in normal use, so the maximization of the soa usage becomes less important, since we put the slopes so high above the normal usage, to keep high headroom.
If there is a good way to not use limiters, I'll prefer using that, if it's good enough.
I want to try to avoid them, but still have good protection. The amps I build, although diy, are used in PA and can be subjected to some abuse, so I want to build something reliable and sturdy. Most of the time though, those amps are used for personal use, so they must be rugged for PA, yet best sounding possible for easy listening in private.
Lucky you!
I haven't blown any speakers or amps either, and I am also quite careful with cabling, so I never shorted anything, but an accident can happen. Better be safe than sorry, and I would hate to blow my nice diy jbl speakers.
That could work if I'm always the one at the volume command on the mix table, but that may not always be the case and I want input muting that also detects input signal level, so the amp doesn't unmute on an overly strong input signal.
I see what you mean. But what I am designing isn't the simple pcb for a diyer to slap into some wood box (including the heatsinks).
I want to make something that I really want to have, and have diy fun at the same time. And it's so much more interesting and educational to research the electronics needed to achieve this. But like I said earlier, I want certain functions, but without making it a kludge.
What I'm seeking is to get the needed function with elegant solutions that aren't overly complex.
I've seen diy builds of amps with lcd displays, with micro-controllers. Very ncie and neat, but I won't go that far.
It's your amp, your choices. You know where I stand.
Yes, that is why I am also looking for other ways to protect besides the vi limiters. And among that, a good clipping prevention, that can double up as an input mute. The clipping prevention would not only prevent clipping, it would also prevent overloads on "normal" loads. Then what's needed is a detection of overload on overly low loads, such as 2ohms or less, and act in a way that prevents going out of soa, without intruding with a vi limiter. One thing planned is temperature sensors on the heatsinks for sure.
Those over-drive and over-load detections can be used to act upon the input signal strength, using the same input limiter for anti-clip and muting.
Some good electronics design work ahead...
In my leach amp sims lately, I've used the ksa1381/ksc3503 for the vas and pre-drivers. Those should suffice.
I do like the specs on the mje15034/5, but it seems they're hard to get.
I have switched to using the mje15032/3 for the drivers.
My ongoing work design right now is aimed at using the mj15003/4 for outputs. But I see they don't perform so well as the mjl4281/4302 in the sims. However those were used by so many in leach amps, I don't see why not using them. I have those in stock, which is one of the main reasons for this current design choice. The plan is to use other types for later designs.
That's why I was considering them earlier, but if they're hard to get, then they're not a workable option.
I know we can get them from onsemi directly, and I did get parts from them several times in the past, however at the present I am not on us soil and the shipping+customs charges are just outrageous, so I won't count this as an option any more unless some other means of transport can be used. (perhaps a group buy some day...)
Your PA application changes everything, I had assumed home use. Full protection is the way to go.
Mouser US and France show the MJE15034/5 in stock. The G sugffix just means RoHS compliant parts. ON Semiconductor MJE15035 Transistors Bipolar - BJT | Mouser
Mouser US and France show the MJE15034/5 in stock. The G sugffix just means RoHS compliant parts. ON Semiconductor MJE15035 Transistors Bipolar - BJT | Mouser
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.