Some time ago we decided to do our own switching poweramp based upon the TC elctronic switchamp module.
This module is also sold by LC Audio, but has nothing to do what so ever with the older ZAPpulse modules. LC Audio was aquired by TC Electronics some years ago, therefor the alternative outlet exists.
By now we reached a point where af few final decissions is to be made regarding design of PSU and the input buffer, but nevertheless some might find the project interesting anyway.
The module from TC was first presented to my by ts designer, who is a master of science who earlier worked @ Power House, which is the home of ICEpower, later he came to TC, where he developed this module.
The very interesting thing about this module is its design of the feedback loops, which I will come back to a bit later.
When I firstlistened to it, it was with a somewhat undersized SMPS and a standard buffer and in addition some safety circuits and ekstra filtering, but in spite of that, it sounded very different from other class D amps I´ve listened to. To me the original ICE modules has a somewhat laid back sound, but endeed they are very clean and smooth.
This one sounds deffinately more neutral to me
So we devided to one ourselves, because its very upside was as mentioned the sound, but also its low energy consumption, which has to be held up against my Gravity class A amp, which consumes 1.2KW and runs @ 55dgr. C and heats up my listning room to around 26 dgr. C on cold winterdays, in the summer not much music is played, since only "nakedlistening" is possible, and we can´t svcare the kids that way - can we?
So class D seems atractive, at least during the summer.
Well! back on track. The special thing in ICE power modules is their patended feedback principple, namely only ICE has 2 different feedback loops. One for high frequency control taking its feedback signal inside the outputfilter, the next loop is an audio frequency loop which takes its feedback signal outside the output filter. This is the very basic thing about ICE, all other makes do only have one feedback loop, which either takes its feedback signal inside OR outside the output filter, mostly inside. That makes the output filter a very critical component, and it leaves the designer with a huge problem, namely a very unlinear output impedance, and it can be hard to obtain good stability.
Also "dead time" adjustment can be critical especially over time.
If you want to know more about the ICE principle you can read more here: ICEpower — Ph.D. thesis
This is the Phd. on which the whole ICEpower business is built.
Looking at ICE power modules and their performance, you´ll find very high powerdensity and pretty nice data generaly, but looking at the output impedance, the ICE modules impedance also rises with frequency, though it has a damping factor @ 1KHz about 1.000 which has fallen to around 20 I think @ ca. 20KHz.
This is due to output filter and its limitation of the speed of the audio feedback loop, but still this is a lot better than amps with only one loop.
The ICE modules are this way superior regarding stability, output impedance and it can handle pretty difficult loads as long as they are not to close to purely capacitive loads.
Now comes the surprising part TC introduced a THIRD feedback loop - yes they did
.
They feature a high speed feedback loop inside the filter, an audio loop aoyside the filter, and in addition a high speed loop outside the filter eliminating the downsides in the filter, which is rising impedance with frequency.
Still the output impedance rises though, but the level is much lower.
The output impedance @ 1KHz is only a few µOhms giving a damping factor that counts in MILLIONS
This is amazing
even @ 20KHz the output impedance is still lower than I think what ever you might imagine, also THD+N is very low, and hardly affected by outputpower.
Unfortunately they filed a patent for this technology, so this is a TC only product.
The sound of this I think will amaze a lot of people, since I´ve never heard anything perform with that ease and overview as this. and for the funny part, it does this in our prototype with an undersized linear powersupply @ 100VA and 18 panasonic FC smooting caps each 660µF. The available peak power is much larger than one imagines though, but that is of course of short duration only. But this really makes one think twice about amps with large caps and monster transformers. In this case something completely different is at stake.
The prototype was actually just meant as an experiment, where we wanted to find out what the module is sensitive to. So we had much larger smoothing caps earlier we tried out RIFA PEH 160, PEH 200, EPCOS Sikorel 105 dgr. and also some Philips and a lot of bypasses.
The thing this module apparently is sensitive to is impedance, and in that way the bunch of FC´s are much lower than any of the other ones, and this actually works much better. We will try out larger values in due time, but certainly it will be 20 or so small caps parallelled and from either Panasonic or Sanyo.
The buffer we made for the prototype is a discrete non feedback unity gain buffer with 2. order one point servo, decoupled with SMR caps from Evox.
We did not yet build an onboard regulator for it, this will come in a later version. For now we just regulate in one stage with 7812/7912 for both the buffer and the modules 12V supply. This is not optimized yet, the module has very high PSRR, but the buffer does not, therefor a high speed series regulator close to the buffer is called for.
So far i really like the switch monster which we named "BIGFOOT".
If you might have experience with this module or good ideas for design details you are welcome to comment, inspiration is highly appreciated.
Also feel free to ask about details regarding the project.
PS!
There is not yet any schematics for the buffer, as it is not yet closed.
This module is also sold by LC Audio, but has nothing to do what so ever with the older ZAPpulse modules. LC Audio was aquired by TC Electronics some years ago, therefor the alternative outlet exists.
By now we reached a point where af few final decissions is to be made regarding design of PSU and the input buffer, but nevertheless some might find the project interesting anyway.
The module from TC was first presented to my by ts designer, who is a master of science who earlier worked @ Power House, which is the home of ICEpower, later he came to TC, where he developed this module.
The very interesting thing about this module is its design of the feedback loops, which I will come back to a bit later.
When I firstlistened to it, it was with a somewhat undersized SMPS and a standard buffer and in addition some safety circuits and ekstra filtering, but in spite of that, it sounded very different from other class D amps I´ve listened to. To me the original ICE modules has a somewhat laid back sound, but endeed they are very clean and smooth.
This one sounds deffinately more neutral to me
So we devided to one ourselves, because its very upside was as mentioned the sound, but also its low energy consumption, which has to be held up against my Gravity class A amp, which consumes 1.2KW and runs @ 55dgr. C and heats up my listning room to around 26 dgr. C on cold winterdays, in the summer not much music is played, since only "nakedlistening" is possible, and we can´t svcare the kids that way - can we?
So class D seems atractive, at least during the summer.
Well! back on track. The special thing in ICE power modules is their patended feedback principple, namely only ICE has 2 different feedback loops. One for high frequency control taking its feedback signal inside the outputfilter, the next loop is an audio frequency loop which takes its feedback signal outside the output filter. This is the very basic thing about ICE, all other makes do only have one feedback loop, which either takes its feedback signal inside OR outside the output filter, mostly inside. That makes the output filter a very critical component, and it leaves the designer with a huge problem, namely a very unlinear output impedance, and it can be hard to obtain good stability.
Also "dead time" adjustment can be critical especially over time.
If you want to know more about the ICE principle you can read more here: ICEpower — Ph.D. thesis
This is the Phd. on which the whole ICEpower business is built.
Looking at ICE power modules and their performance, you´ll find very high powerdensity and pretty nice data generaly, but looking at the output impedance, the ICE modules impedance also rises with frequency, though it has a damping factor @ 1KHz about 1.000 which has fallen to around 20 I think @ ca. 20KHz.
This is due to output filter and its limitation of the speed of the audio feedback loop, but still this is a lot better than amps with only one loop.
The ICE modules are this way superior regarding stability, output impedance and it can handle pretty difficult loads as long as they are not to close to purely capacitive loads.
Now comes the surprising part
TC introduced a THIRD feedback loop - yes they did.They feature a high speed feedback loop inside the filter, an audio loop aoyside the filter, and in addition a high speed loop outside the filter eliminating the downsides in the filter, which is rising impedance with frequency.
Still the output impedance rises though, but the level is much lower.
The output impedance @ 1KHz is only a few µOhms giving a damping factor that counts in MILLIONS
This is amazing
even @ 20KHz the output impedance is still lower than I think what ever you might imagine, also THD+N is very low, and hardly affected by outputpower.Unfortunately they filed a patent for this technology, so this is a TC only product.
The sound of this I think will amaze a lot of people, since I´ve never heard anything perform with that ease and overview as this. and for the funny part, it does this in our prototype with an undersized linear powersupply @ 100VA and 18 panasonic FC smooting caps each 660µF. The available peak power is much larger than one imagines though, but that is of course of short duration only. But this really makes one think twice about amps with large caps and monster transformers. In this case something completely different is at stake.
The prototype was actually just meant as an experiment, where we wanted to find out what the module is sensitive to. So we had much larger smoothing caps earlier we tried out RIFA PEH 160, PEH 200, EPCOS Sikorel 105 dgr. and also some Philips and a lot of bypasses.
The thing this module apparently is sensitive to is impedance, and in that way the bunch of FC´s are much lower than any of the other ones, and this actually works much better. We will try out larger values in due time, but certainly it will be 20 or so small caps parallelled and from either Panasonic or Sanyo.
The buffer we made for the prototype is a discrete non feedback unity gain buffer with 2. order one point servo, decoupled with SMR caps from Evox.
We did not yet build an onboard regulator for it, this will come in a later version. For now we just regulate in one stage with 7812/7912 for both the buffer and the modules 12V supply. This is not optimized yet, the module has very high PSRR, but the buffer does not, therefor a high speed series regulator close to the buffer is called for.
So far i really like the switch monster which we named "BIGFOOT"
.If you might have experience with this module or good ideas for design details you are welcome to comment, inspiration is highly appreciated.
Also feel free to ask about details regarding the project.
PS!
There is not yet any schematics for the buffer, as it is not yet closed.
Last edited:
yeah!
ICE power has a very nice bass control, due to the low output impedance, but so does my class A amp, which features 16 matched output transistors pr. ch. Which results in a dampingfactor @ around 400 from 20-20 KHz.
This is easily heard as a very controlled behaviour no matter what is to be done throughout the audible frequency area.
But this small fellow is somewhat of a beast in that regard
We still need to try out a few things regarding PSU´s, but already now I can say, that it really has its moments.
Compared to ICE modules, I think the TC is sounding clearer and more lifelike.
Hi Kurt von Kubik
Sounds interesting. The output filter impedance is definitely one hurdle to overcome and it sounds like TC has an interesting way of controlling the output filter impedance curve.
Kurt, what's the PSRR of this amp?
Does this amp have low reactivity to speaker load? If so, is it due to the implementation of the 3rd feedback loop?
Sounds interesting. The output filter impedance is definitely one hurdle to overcome and it sounds like TC has an interesting way of controlling the output filter impedance curve.
Kurt, what's the PSRR of this amp?
Does this amp have low reactivity to speaker load? If so, is it due to the implementation of the 3rd feedback loop?
Hi Kurt von Kubik
Sounds interesting. The output filter impedance is definitely one hurdle to overcome and it sounds like TC has an interesting way of controlling the output filter impedance curve.
Kurt, what's the PSRR of this amp?
Does this amp have low reactivity to speaker load? If so, is it due to the implementation of the 3rd feedback loop?
Hi KLe!
I also found the TC way of solving the output impedance problem sofisticated.
It is not so that they made it doing "clean and smooth" measurement results, but they did manage to get both THD+N and output impedance down to almost unimagenable values.
The PSRR is not stted by TC, but we,´ve made experiments with very high ESR caps from Philips, resulting in ripple @ 5V and even more, without any influence on power delivery. This is due to the design, which if the supply voltage decreases, then the modulator will just increase time. So theoretically ripple is no issue for concern.
But a sitching amp do need power en very short jerks, wich makes the high frequency behaviour of the power supply pretty important. Therefor a very low impedance power supply is needed, and at best if it has low impedance at high frequencies as well.
The abillityy to drive difficult speakers speakers should be excellent, though I do not know how it responds on reactive loads. reactive loads is normally a challenge for switching amps. I´ve also seen some trouble with switching amps, that cannot be turned on without load, I think I´ve seen 3 or 4 switcing amps send out smoke doing that.
This one is perfectly stable in any regard, even if you pull the plug playing really loud, you just plug it again, the only thing that happans is: music - no music and so on. No blob bump fssst or anything at all.
TC states its max output current to more than 60A peak, and max power is 200 Watts @ 8 Ohm, DNR is a smashing 113 dB with standard buffer unweighted.
It is a very strong amplifier indeed, and I think it must be due to the third loop, because it lowers the outputimpedance that much, that is my belief.
The only special thing about this module is, that it requires an input buffer, because of a very low input impedance @ 2K Ohms. That will not go with most preamps without change in sound.
are we talking now about the current LCaudio Class D module?
Is it available yet?
What are the actual technical data?
how much current (rms) caoul you get out of them?
I mean are they capable of driving 4 ohm loads in Bridge mode?
and more likely, are they now (if I read your post correctly) using post filter feedback? (e.g. the frequency response is load independent??)
Thanks,
Tamas Tako
Yes it is the current LC module, but it is not designed by Lars Clausen.
Have a look here L C Audio Technology / Forside
Data
Standard setup med 500VA toroid og PredatorXE supply output power 8 Ohm 230 Watt RMS
Bridged output power 4 Ohm 1200 Watt RMS
Peakcurrent >60 Ampere
Frequency response 20-20khz +/- 0.2dB
THD+N, 20-20khz, 1W i 8 ohms < 0,004%
THD+N, 20-20khz, 100W i 8 ohm < 0,1%
THD+N, 20-20khz, 250W i 4 ohm < 0,1%
Gain with buffer 35dB
DNR, with buffer, uvejet 113dB
Indputimpedance with buffer 10 kOhm
Without buffer buffer 2 kOhm
Supply voltage +/-48-70 Volt DC
Efficiency, 300W i 4 ohm ~93%
About the feedback, they are using pre AND post filter FB, 3 loops with different properties in conjunction.
The modules are available for DIYérs, LC Audio is the outlet.
Hi,
is this 60A peak = 42.5Arms????
meaning the max Prms to be 42A x 42A x 4Ohm in bridge?? (=7700W) meaning, that just the DC voltage rails are the limiting Factor? (70V/1.4142 x 2= ca 100V in bridge mode =2500W rms??)
i highly doubt these 60A peak seeng just 1 TO220 device there per switch...
Thanks,
Tamas
Well!
I do only have the data provided from LC Audio, so I do not really have any Idea of under what conditions the peak current is available.
But I talked to the designer, and he also mentioned 60A peak current.
About the two TO220 FETs in the switch, one must understand, that the switching principle completely eliminates the need for dozens of transistors and bulky heatsinks.
In linear design you can gain sound quality by oversizing both your amp and the PSU, therefor measurement condition for linear amps is often much harder than any condition they will face in real life. I.e. for class A/B amps, 1/3 power for one hour @ 1.000Hz sine wave is the preconditioning routine @ stereophile, which is the worst possible thermal condition.
If the amp copes with that, it is correctly sized and it will normally behave both strong and with authority in real life situations.
Switching amps behaves completely different, you´ll win absolutely nothing by oversizing it to meet the conditions mentioned above. The point is that the switch does not perform significantly worse @ high power than @ low power delivery, hence th normal oversizing does not make any sense in this case.
How much power it can deliver on long term basis is very dependend on cooling, since the remaining 7% not delivered to the speaker, discipates as heat in the switches. But careful thermal design will indeed make it very strong.
Actually there is a significant degradation of switching performance, when an amp enters continous inductor current mode and this gets worse with inreasing current.
But I agree that oversizing is not needed nor desirable, because switching mosfets work fine with their square SOA.
Yes - Right you are!
But not with this one, and that is due to the 3 different feedback loops, as mentioned in the first post.
The only thing I think it will not like, is purely capacitive loads, as it is not designed to feed anything like that.
It will deliver 60Amps peak! Only for short periods, but still... It does 60Amps.
The module is rated at 230 watt@8ohms continuously. My guess is 450 watt@4ohms.
We have been playing on 4 ohm speakers at levels where we need to leave the room, and still the temperature of the module is less than 30 degree C.
I think you are looking for a non-existing problem. We have not yet found any music, that will the amp more than 30 degree C, when mounted on a small piece of aluminum (No heatsink). And trust me... we did try!
Last edited:
Hey Kle!
It is the current module incorporating the 3 feedvack loops, I do not really know if there was an earlier module, since I´ve never really had any interest in LC gear.
It was first when the designer showed me the current module and told about the peticulars about it, and also made me listen to it, that I became interested in the technology.
Originally he was asked to look at the LC modules, to see if they could also be incorporated in some of TC´s pro-gear i.e. studiomonitors, guitar amps, and Lab Gruppens PA amps and so on. But they were discarded all of them, because of lack of stability amongst other things.
So he had to start of from scratch, and I think the result is impressing, at least when a high quality buffer and good suplies are used.
Hi,
this module is good but ...I think that is not new technology. 3 FB are good and not easy to do, one for digital to digital to adjust the stretch near the rail voltage (PWM) and the other two (when you can do) fixed a few little things .. including the filter(Toroidal) response at audio frequencies.
Problem is PSU for amplifier not for ..lamp!
Best Regards
this module is good but ...I think that is not new technology. 3 FB are good and not easy to do, one for digital to digital to adjust the stretch near the rail voltage (PWM) and the other two (when you can do) fixed a few little things .. including the filter(Toroidal) response at audio frequencies.
Problem is PSU for amplifier not for ..lamp!
Best Regards
Whether this is new technology or not, I do not really know.
But the fact is, that 2 feedback loops was patended by B&O and is still the foundation of ICE power, which by far is the most powerfull and stable class D for general purpose.
The TC takes this a bit further, by incorporating the 3.rd loop, which provides dampingfactor hundreds of times larger than all other class D amps, and also THD + N is pretty impressing.
The powersupplies though are IMHO pretty much a matter to consider, just as well as a really good buffer is.
Btw. TC patended the technology, which might be an indicator of first time ever.
The two feedback loops on ICEpower are redundant. The same result may be achieved with a single post-filter loop and an output filter and layout having low parasitics. That's UcD, although it may be considered as having two loops because the signal taken from the output follows two different paths. In general, I think that a "single" feedback loop (like that) can do it all, but it may include several poles.
The 60A peak current capabilty may easily be real, although that's not very reliable for one pair of MOSFET with such a small heatsink, unless there is time/temperature dependent current limiting. But then the modules should also be rated at 2 ohm, and they aren't. Also, output coils don't look like capable of handling more than 10A rms or so for some time.
For example, my current project can do around 5kw on 2 ohm for 1 second (with time-dependent true-RMS current limiting to protect not only the amplifier but the power source and the load ). Current limit is 75A peak and 25A rms long term, but that's achieved using pairs of parallel MOSFET, and 13 AWG wire on output coils. There are low cost tricks to calculate rms current without expensive multiplier ICs.
The 60A peak current capabilty may easily be real, although that's not very reliable for one pair of MOSFET with such a small heatsink, unless there is time/temperature dependent current limiting. But then the modules should also be rated at 2 ohm, and they aren't. Also, output coils don't look like capable of handling more than 10A rms or so for some time.
For example, my current project can do around 5kw on 2 ohm for 1 second (with time-dependent true-RMS current limiting to protect not only the amplifier but the power source and the load
). Current limit is 75A peak and 25A rms long term, but that's achieved using pairs of parallel MOSFET, and 13 AWG wire on output coils. There are low cost tricks to calculate rms current without expensive multiplier ICs.
Last edited:
The two feedback loops on ICEpower are redundant. The same result may be achieved with a single post-filter loop and an output filter and layout having low parasitics. That's UcD, although it may be considered as having two loops because the signal taken from the output follows two different paths. In general, I think that a "single" feedback loop (like that) can do it all, but it may include several poles.
The 60A peak current capabilty may easily be real, although that's not very reliable for one pair of MOSFET with such a small heatsink, unless there is time/temperature dependent current limiting. But then the modules should also be rated at 2 ohm, and they aren't. Also, output coils don't look like capable of handling more than 10A rms or so for some time.
For example, my current project can do around 5kw on 2 ohm for 1 second (with time-dependent true-RMS current limiting to protect not only the amplifier but the power source and the load
Whether the modules can handle the power specified or not, I do not know, since I did not measure it, but I have no reason to doubt it what so ever. They are indeed very strong.
About the multiple feedback loops vs. only one, then this is THE item in class D amplification.
A feedbackloop around the LPF filter in the output will always be slow, and therefor be audio frequencies only. This given the output impedance will rise with frequency, typical the amping factor ends up around 1-10 @ 20KHz or even worse.
In addition stability is poor.
ICE Power with its extra loop, is both very stable and also has a smoother decrease in dampingfactor than alternatives.
The TC module though has an outputimpedance @ 1KHz measuring a few µ Ohms, giving it several millions in dampingfactor, of which tens of thousand are still available @ higher frequencies.
When listening to it, this is excactly what comes to mind, since it starts and stops completely undramatic and low level information is stunning.
My other amplifier is a huge class A amp rated @ 2*100 Watts of class A power @ 8 Ohms. It can deliver 60A for ever p.ch , and can with easy drive ½ Ohm speakers (of course this would not be class A anylonger). The dampingfactor is around 400 even though it is a non feedback design. So this is a pretty strong amp. But the TC does not seem any weaker, there is sonic differences though, but the strengt and authority of these amps are of comparable magnitudes.
That is what originally impressed me about this module, later on performance has been bettered due to extra attention payed to both supplies and input buffer.
I think TC somewhere has a more detailed datasheet on the module, than the specs supplied by their LC outlet, but unfortunately I cannot seem to find it anywhere.
Hello, I agree with EVA for FB on ICEpower.
Let me clarify a point about class D amplifiers
500w to Kw for commercial audio, requires to solve several problems to class D for audiophile performance.
I concentrated my resources to design new systems of protection for high power, short cut at max power, and innovative clip-control, when audio envelope force HI-current.(I not decrease current but audio signal) in realtime.
I find it fascinating at the time resolve some intrinsic defects of class D (as compared to professional class AB for audiophile). .. former jazz sound reflected lower resolution at high frequencies, voice over 70% of pwm ... contains many harmonics and sound is not pure. raising the percentage of PWM, dramatically increases the IMD and finally, when suddenly we are on high efficiency. what good is power after 80% of the PWM if the sound sucks on this section? (thermal has already won).
-----
For SMPS, all in the market, if carefully measured the amp .. I would take it and run it from windows (to be honest) only for high power are good.
a company is launching a new power supply on the market, truly revolutionary. no inductors and toroids, tested on mospower class AB, delivers 500W RMS (4R) without ripple and EMI.
see audiopower.it
I learned that is based on a new technology "RIPS"(Rotary ignition Power system), the generator current is modulated by the current from amplifier up to 100kHz.
I asked for a DPS-600 Amplifier to measure it ... let's see what happens.
Let me clarify a point about class D amplifiers
500w to Kw for commercial audio, requires to solve several problems to class D for audiophile performance.
I concentrated my resources to design new systems of protection for high power, short cut at max power, and innovative clip-control, when audio envelope force HI-current.(I not decrease current but audio signal) in realtime.
I find it fascinating at the time resolve some intrinsic defects of class D (as compared to professional class AB for audiophile). .. former jazz sound reflected lower resolution at high frequencies, voice over 70% of pwm ... contains many harmonics and sound is not pure. raising the percentage of PWM, dramatically increases the IMD and finally, when suddenly we are on high efficiency. what good is power after 80% of the PWM if the sound sucks on this section? (thermal has already won).
-----
For SMPS, all in the market, if carefully measured the amp .. I would take it and run it from windows (to be honest) only for high power are good.
a company is launching a new power supply on the market, truly revolutionary. no inductors and toroids, tested on mospower class AB, delivers 500W RMS (4R) without ripple and EMI.
see audiopower.it
I learned that is based on a new technology "RIPS"(Rotary ignition Power system), the generator current is modulated by the current from amplifier up to 100kHz.
I asked for a DPS-600 Amplifier to measure it ... let's see what happens.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.