What to do. OK, your tweeter will be at least 3 dB , probably much more efficient that your woofer. So, in a passive crossover you would have had an L pad anyway. Just put a series 4 or 6 Ohm resistor in line with it (with the DC blocking cap). Then you are safe as well as reduced the distortion.
For the woofer, if you are stuck with such a bad choice, you could limit the bandwidth of the amp quite a bit and adjust the compensation. Realize the current from the transformer has to be much more. Where I need 2.5 A into 8 Ohms, you now need 10 amps for the same voltage out. To then keep it within the SOA, you need to reduce the voltage a lot. A single pair looks OK at about 40 Watts with 36V rails. With 12V rails, you pull 3.5 Amps into 3 Ohms for 40W. I bought 3 Amp 12V transformers PER SIDE. I am designing for 8 Ohm nominal, 6.1 Ohm DC RE, and expect 5 Ohm impedance speakers and going the extra mile to be within SOA with 4 Ohm load and not self destruct with instability into 3. I have owned unstable amps before, not again.
We don't have a data sheet for the Chinese outputs, so we can't do a proper SOA analysis anyway.
For the woofer, if you are stuck with such a bad choice, you could limit the bandwidth of the amp quite a bit and adjust the compensation. Realize the current from the transformer has to be much more. Where I need 2.5 A into 8 Ohms, you now need 10 amps for the same voltage out. To then keep it within the SOA, you need to reduce the voltage a lot. A single pair looks OK at about 40 Watts with 36V rails. With 12V rails, you pull 3.5 Amps into 3 Ohms for 40W. I bought 3 Amp 12V transformers PER SIDE. I am designing for 8 Ohm nominal, 6.1 Ohm DC RE, and expect 5 Ohm impedance speakers and going the extra mile to be within SOA with 4 Ohm load and not self destruct with instability into 3. I have owned unstable amps before, not again.
We don't have a data sheet for the Chinese outputs, so we can't do a proper SOA analysis anyway.
That's true. but you can use an equivalent type and set the SOAR limits 10-20% less.
If the transistors blow up in that setting, you know you had to buy decent transistors in the first place.
What? I can't believe this rubbish... "no amp is a good choice for 4 ohms". You should not be posting on this web site if these are the kind of blanket statements you make, or you should at least provide more detail to back up what you are saying. As they stand they are completely misleading. There are plenty of class AB amps out there that can drive 4 ohms loads or less just fine. Sure you might need more output pairs to keep up with the current demand, but that is nothing new.
Last edited:
OK, you are certainly entitled to your opinion. I didn't quite see that you were intending to mean "in my opinion..." by your word choices. It seemed to me that you were being much more definite than that...
Can you make it clearer WHY you feel that
What in your opinion is so difficult about driving a 2 ohm (resistive) load (in general, not with 2 devices) with a class-AB amp? It's just a matter of current supply and demand, isn't it? We have these lovely VMOSFETs available these days, so thermal runaway can be designed away. What else? I'm now asking this in earnest, because I am not an amp designer.
All you need to do is model it. Two big problems, ability to main stability and greatly rising distortion. From a systems design standpoint, it is far better to have a higher impedance load. I was not aware of this before I started studying amplifier design about a year ago. Read Self and Cordel as a good start. I knew I was not happy with smaller amps on speakers with low impedance, but did not understand the why. I think my next MMT speaker will be 12 Ohms.
In another 10 years or so, I may understand it well enough to say I am competent to design an amp, but in the mean time I have good information that will improve my speaker designs. Knowing what not to do is a good start.
My learning project I am calling the "Reasonable" amp. I have several iterations with different output stages so I can build several with only the output change to hear what the subjective differences really are. L-fet drivers, bjt outputs, BJT drivers, Lateral or HEXFET outputs etc. All EF so easier to keep stable with multiple pairs. I have owned many amps and can ascribe various attributes to them, but I can't say I can attribute anything to the technology. Could just be execution. You can only learn so much from books and sims. Build and listen is next.
The MX-50 is a single pair CFP BJT output. The only thing I need to make a complete amp is a $20 transformer, so it is a fun budget build I can use to test the stability in silicon to the simulation. It's distortion as sim is lower than I can probably measure but that is another reality check to follow through on. I may make the famous distortion multiplier to see if that helps.
In another 10 years or so, I may understand it well enough to say I am competent to design an amp, but in the mean time I have good information that will improve my speaker designs. Knowing what not to do is a good start.
My learning project I am calling the "Reasonable" amp. I have several iterations with different output stages so I can build several with only the output change to hear what the subjective differences really are. L-fet drivers, bjt outputs, BJT drivers, Lateral or HEXFET outputs etc. All EF so easier to keep stable with multiple pairs. I have owned many amps and can ascribe various attributes to them, but I can't say I can attribute anything to the technology. Could just be execution. You can only learn so much from books and sims. Build and listen is next.
The MX-50 is a single pair CFP BJT output. The only thing I need to make a complete amp is a $20 transformer, so it is a fun budget build I can use to test the stability in silicon to the simulation. It's distortion as sim is lower than I can probably measure but that is another reality check to follow through on. I may make the famous distortion multiplier to see if that helps.
Problem with 2 ohm loads is the current requirements of the output stage.
E.g.: Amplifier railvoltage (+ and -) : 20 Volts
With a 2 ohms load this would require 20v/2 Ohms = 10 Amps current capability of the output stage.
Now look at the SOAR of a populair power transistor e.g 2sc5200:
It's max current is 15A. But at 20 volts, only 7 amps remain... (dc load!). 100ms current capability is approx. 15 Amps. So one transistor will cover our requirements.
When using higher voltages current capability of the transistor falls, cause the max. power rating can not be exceeded. So, at 30 volts only 5A (dc) resp. 9.5 A(100ms) remain... To guaranty stable operation at 30Volts and 2 ohms load two transistors are required!
At 45V rail voltage stable (2 Ohm) operation requires 4 output transistors (100ms graph), but yields no reserve!
I guess this is why Tvrgeek says that 4 ohms speakers are a bad load... for current requirements are twice as high!
Further: observe that the MX-50 amplifier has no protection circuit for the power transistors. So using 4 ohm speakers (with a rail voltage of 30 volts) might ruin your power stage when driven hard.
Last edited:
I am working on some protection. The traditional output I/V takes too many parts and causes big issues when it activates. Baker clamp and flying diodes on the bias, and I am looking at how some behave in other places. Keeping the outputs off the rails is harder than one may think.
12V looks like the optimum rails for the MX-50.
I also think my guesswork, trial and error, on the TMC is not optimum. More work to do. The biggest thing I am fighting is the glitching as the drivers turn off. This does not directly effect the outputs as they are off at that time, but they get reflected back through the circuit and cause ugly ringing in various untoward places. This may be just how CFP works, and it may be irrelevant. Until I get the transformers in and build on e on the bench, I don;t know.
12V looks like the optimum rails for the MX-50.
I also think my guesswork, trial and error, on the TMC is not optimum. More work to do. The biggest thing I am fighting is the glitching as the drivers turn off. This does not directly effect the outputs as they are off at that time, but they get reflected back through the circuit and cause ugly ringing in various untoward places. This may be just how CFP works, and it may be irrelevant. Until I get the transformers in and build on e on the bench, I don;t know.
Tvrgeek: why are you trying so hard to drive 2ohm loads? A decently designed speaker never dips so low even if its 4 ohm. If it does, it's not properly designed. A quick impedance measurement can do wonders... Also a decent speaker reaches 100dB or more with the first let's say 10-15watts... Or are we going to amplify on stage?
baskiria,
You hit the nail right on the head! A DECENT speaker never dips below 4 Ohms. The OP was considering using this amp to drive a 4 Ohm nominal driver. It will have a DC resistance of 3.6 or so and a complex impedance likely closer to 2.5. 4 Ohms is a great target considering most mid-woofers out there are about 6.5 Ohms DC. Why would I pick 2 if I am pretty confident I will never go below 4? Safety margin.
In the PA world, we have other options. Any competent sound engineer will have compressors and peak limiters. These days, mostly all digital amps. They will have measured and profiled the system so they can stay out of trouble. Having your lead vocal amp blow up in a concert is considered very bad form. Notice I did say "competent". I will still never forgive the Moodey Blues for their totally incompetent tour engineers in the 70's.
One thing studying amps has taught me is to NOT drive low Ohm loads, so I will be careful that my designs don't go too low. The 12V version I am building will be driving my own speaker so I will select the stability to be suitable.
Should all amps be at least stable at 2 Ohms? I argue yes. You don't know anything about the speaker if you did not design it. Having owned a commercial amp with insufficient stability margin, I do not intend to go there again. Some unconventional speakers and worse, snake oil speaker cables, can make things even worse. Look at the history of Maggies and Apogee's as examples of bad loads known to put esoteric amps into instability. (esoteric, as in badly designed, expensive, and raved about in magazines as opposed to big expensive well designed amplifiers)
Music has transients. Even though you may be averaging 3 or 4 Watts, you need to have the overhead not to clip, or to do so cleanly for transients that can be much larger. It is suggested this is why the tube club gets away with small amps, they clip very cleanly, where a SS amp does nasty things that take time to settle out. Behavior of the amp over low distortion numbers? I am rapidly being led down that path. I have heard amplifiers with published THD in the .000x range that sounded terrible, and ones in the .1 range that sounded wonderful. I have also heard amps in the .0x range that were better still. I still believe harmonic distortion is the issue, just I now believe it is the transient conditions where we have the problems. .05 under any condition as a reasonable target?
So, for the MX-50-TVR I am using 12V rails for about a 25W amp. I had originally planned on 22V rails. I am not too worried about SOA and I am trying to implement protection to prevent the outputs from reaching the rails. That looks like the sweet spot for this specific amplifier design. My "Reasonable" design I am targeting 45W with 36V rails. It is an EF with two pairs of outputs.
Now, there is a completely different concept. High voltage rails, not enough current, big supply caps and hope you never clip, e.g. the MX-50 60W as advertised with 45V rails. I can accept that, but one had better implement hard I/V limiting. (which usually sounds very bad). There si no room on these little boards to *** the extra transistors needed. This thread is working within the constraint of the MX-50. This is how most consumer amps are built today. How else do you think they can put seven 120W amps into a 160mm high AVR chassis? IMHO they sound like crap.
A quick impedance measurement of the speaker does not show the effective load considering the phase. It does not consider back EMF. You need a full Ohm lower if setting SOA and stability on a restive load. I am trying to come up with a SPICE model for a bad load, starting with Cordell's nice typical speaker load. I have not gotten the TIAN probe to work correctly, so this will take a lot more work.
You hit the nail right on the head! A DECENT speaker never dips below 4 Ohms. The OP was considering using this amp to drive a 4 Ohm nominal driver. It will have a DC resistance of 3.6 or so and a complex impedance likely closer to 2.5. 4 Ohms is a great target considering most mid-woofers out there are about 6.5 Ohms DC. Why would I pick 2 if I am pretty confident I will never go below 4? Safety margin.
In the PA world, we have other options. Any competent sound engineer will have compressors and peak limiters. These days, mostly all digital amps. They will have measured and profiled the system so they can stay out of trouble. Having your lead vocal amp blow up in a concert is considered very bad form. Notice I did say "competent". I will still never forgive the Moodey Blues for their totally incompetent tour engineers in the 70's.
One thing studying amps has taught me is to NOT drive low Ohm loads, so I will be careful that my designs don't go too low. The 12V version I am building will be driving my own speaker so I will select the stability to be suitable.
Should all amps be at least stable at 2 Ohms? I argue yes. You don't know anything about the speaker if you did not design it. Having owned a commercial amp with insufficient stability margin, I do not intend to go there again. Some unconventional speakers and worse, snake oil speaker cables, can make things even worse. Look at the history of Maggies and Apogee's as examples of bad loads known to put esoteric amps into instability. (esoteric, as in badly designed, expensive, and raved about in magazines as opposed to big expensive well designed amplifiers)
Music has transients. Even though you may be averaging 3 or 4 Watts, you need to have the overhead not to clip, or to do so cleanly for transients that can be much larger. It is suggested this is why the tube club gets away with small amps, they clip very cleanly, where a SS amp does nasty things that take time to settle out. Behavior of the amp over low distortion numbers? I am rapidly being led down that path. I have heard amplifiers with published THD in the .000x range that sounded terrible, and ones in the .1 range that sounded wonderful. I have also heard amps in the .0x range that were better still. I still believe harmonic distortion is the issue, just I now believe it is the transient conditions where we have the problems. .05 under any condition as a reasonable target?
So, for the MX-50-TVR I am using 12V rails for about a 25W amp. I had originally planned on 22V rails. I am not too worried about SOA and I am trying to implement protection to prevent the outputs from reaching the rails. That looks like the sweet spot for this specific amplifier design. My "Reasonable" design I am targeting 45W with 36V rails. It is an EF with two pairs of outputs.
Now, there is a completely different concept. High voltage rails, not enough current, big supply caps and hope you never clip, e.g. the MX-50 60W as advertised with 45V rails. I can accept that, but one had better implement hard I/V limiting. (which usually sounds very bad). There si no room on these little boards to *** the extra transistors needed. This thread is working within the constraint of the MX-50. This is how most consumer amps are built today. How else do you think they can put seven 120W amps into a 160mm high AVR chassis? IMHO they sound like crap.
A quick impedance measurement of the speaker does not show the effective load considering the phase. It does not consider back EMF. You need a full Ohm lower if setting SOA and stability on a restive load. I am trying to come up with a SPICE model for a bad load, starting with Cordell's nice typical speaker load. I have not gotten the TIAN probe to work correctly, so this will take a lot more work.
"Can someone chime in about what rails would be suitable for a 4-ohm nominal load (amp connected to driver, no passive crossover) using the MX50-SE?"
If you use the CFP on the board to drive several pairs of EF outputs you will have an output stage that looks similar to many McIntosh, or an Ampzilla III, and you can drive a 1Ω load (with a suitable number of output pairs and a big VA transformer).
If you use the CFP on the board to drive several pairs of EF outputs you will have an output stage that looks similar to many McIntosh, or an Ampzilla III, and you can drive a 1Ω load (with a suitable number of output pairs and a big VA transformer).
Yea you could. I would not. That would explain why the amps you mention were not very pleasing amps to listen to.
I will say one more time: This amplifier is not well suited for a 4 Ohm nominal load as is. If you must, do the SOA graphs. I am guessing something around 20 volts as long as you don't mind 1% THD.
Never, never, never connect an amp to a tweeter without a blocking cap. I have not built many speakers that don't need some passive filtering, even if it is not strictly speaking a crossover.
I will say one more time: This amplifier is not well suited for a 4 Ohm nominal load as is. If you must, do the SOA graphs. I am guessing something around 20 volts as long as you don't mind 1% THD.
Never, never, never connect an amp to a tweeter without a blocking cap. I have not built many speakers that don't need some passive filtering, even if it is not strictly speaking a crossover.
There are several ways to learn about these things. Go ahead and hook it up any way you think will do what you want. The worst it will do is blow up. When I was younger, I blew plenty of things up because I did not understand it well enough and thought I knew better. Those lessons tend to stick and with an amp, you probably won't get hurt.
Now I am old and cheap, I try to learn from others mistakes. I apply my background in electronics to do the basic analysis so I don't make too many more big mistakes. I use tools like scratch paper and a calculator. Now I have been learning LTSpice which makes it much quicker. Calculators are really cool. They sure beat my slide rule I went to school with.
I am building my MX-50's as a 6.5Watt amps in a mono powered speaker with 3 Amp 12V secondary transformers. The woofer has a 6.7 Ohm Re, so in the sealed cabinet, It should never go below 5.5 I will be adding a series resistor to the tweeter of a couple of ohms along with it's blocking cap. I can thus move the notch filter from the woofer to the crossover helping to reduce a low ohms dip it has as a passive crossover. This is a low power system but of high fidelity replacing a 2.5W class A amp of very low quality.
Your choice. Just remember to have fun.
Now I am old and cheap, I try to learn from others mistakes. I apply my background in electronics to do the basic analysis so I don't make too many more big mistakes. I use tools like scratch paper and a calculator. Now I have been learning LTSpice which makes it much quicker. Calculators are really cool. They sure beat my slide rule I went to school with.
I am building my MX-50's as a 6.5Watt amps in a mono powered speaker with 3 Amp 12V secondary transformers. The woofer has a 6.7 Ohm Re, so in the sealed cabinet, It should never go below 5.5 I will be adding a series resistor to the tweeter of a couple of ohms along with it's blocking cap. I can thus move the notch filter from the woofer to the crossover helping to reduce a low ohms dip it has as a passive crossover. This is a low power system but of high fidelity replacing a 2.5W class A amp of very low quality.
Your choice. Just remember to have fun.
I guess this amp is off the table for me. I do not want to start to learn amplifier design and watch things catch fire as a "learn about" the topic. I really don't give 2 cents about load lines and transistor operating parameters. Honestly, I only have so much time to spend on DIY audio and I am already trying to learn lots in other areas of this hobby. For many things you just have to cut to the chase - for me amplification is one of these.
You should get the MX50X2 then, that adds two more output transistors per side but it's not that dirt cheap.
My opinion however is the first watt counts. I think I hardly go over a few watts or let's be generous and say 10W with normal in room listening. That means 95% of the time casual listening volume that you might need to talk with your voice raised somewhat, and the rest 5% when I go louder for a track or so. And I believe this is what most people do whoever lives in a flat with neighbours around.
I doubt if 4 amps per side is not enough for that. I easily popped a 4amp fuse with a volume knob setting that I never even once tried to reach with normal listening. It was the fuse, and not the output.
And at the end what really matters is how it sounds, and I tell you, it sounds damn good, being cheap is a nice extra.
My opinion however is the first watt counts. I think I hardly go over a few watts or let's be generous and say 10W with normal in room listening. That means 95% of the time casual listening volume that you might need to talk with your voice raised somewhat, and the rest 5% when I go louder for a track or so. And I believe this is what most people do whoever lives in a flat with neighbours around.
I doubt if 4 amps per side is not enough for that. I easily popped a 4amp fuse with a volume knob setting that I never even once tried to reach with normal listening. It was the fuse, and not the output.
And at the end what really matters is how it sounds, and I tell you, it sounds damn good, being cheap is a nice extra.
X2 is a much better choice. Gain clone?
The lesson to take away from here Charlie is that a sound system is a SYSTEM. You need to appreciate how each part works with the next. Hint, a 4 Ohm driver, excluding the special case of a sub, is not intended to be use by itself. It is intended for an MTM or MMT configuration as by itself, puts too much demand on the amplifier. Notice most companies make their small mid-bass in both 8 and 4 Ohm versions. Just like the craze for passive preamps for a while ignored the distortion caused by the loading of the input stage. Did it fail to work? No. Did anyone get hurt? No. But a buffer sounded much better. Much to learn. More than a lifetime. What is depressing, or actually inexcusable, is how much stuff is sold by companies who actually employ engineers that do not take these things into consideration.
The CFP output makes it hard to add a second pair and remain stable, otherwise that would be the easy fix.
My main interest is speakers, but in trying to understand how to design a better speaker, I had to understand amplifiers because they are working as a system. What little I have learned so far has greatly influenced how I am approaching my speaker design.
Best of luck.
The lesson to take away from here Charlie is that a sound system is a SYSTEM. You need to appreciate how each part works with the next. Hint, a 4 Ohm driver, excluding the special case of a sub, is not intended to be use by itself. It is intended for an MTM or MMT configuration as by itself, puts too much demand on the amplifier. Notice most companies make their small mid-bass in both 8 and 4 Ohm versions. Just like the craze for passive preamps for a while ignored the distortion caused by the loading of the input stage. Did it fail to work? No. Did anyone get hurt? No. But a buffer sounded much better. Much to learn. More than a lifetime. What is depressing, or actually inexcusable, is how much stuff is sold by companies who actually employ engineers that do not take these things into consideration.
The CFP output makes it hard to add a second pair and remain stable, otherwise that would be the easy fix.
My main interest is speakers, but in trying to understand how to design a better speaker, I had to understand amplifiers because they are working as a system. What little I have learned so far has greatly influenced how I am approaching my speaker design.
Best of luck.
"That would explain why the amps you mention were not very pleasing amps to listen to. "
Confirmation bias?
McIntosh makes three amplifiers that have the exact same driver card, and either one, two, or three pair of outputs. The one with two pair of outputs sounds the best.
I doubt you ever heard an Ampzilla III, it failed as a commercial venture, and none were ever produced (unless you saw the schematic and built one from scratch).
So what do you think of the Bryston two pair output stage, one pair common collector, the other pair common emitter? Have you heard one?
Confirmation bias?
McIntosh makes three amplifiers that have the exact same driver card, and either one, two, or three pair of outputs. The one with two pair of outputs sounds the best.
I doubt you ever heard an Ampzilla III, it failed as a commercial venture, and none were ever produced (unless you saw the schematic and built one from scratch).
So what do you think of the Bryston two pair output stage, one pair common collector, the other pair common emitter? Have you heard one?
There is. On a CFP, it is the drivers you monitor. Both drivers share the heat sink with the bias servo. Correctly done.