Hi, FAB,
Hi, Mike,
Yes, the ordinary configuration is like what you make. But somehow IGBT is "mosfet like" gate/base and "bipolar like" C-E/D-S. The advantage of mosfet in front that it will "block" the final stage from the previous driver stages (strange or heavy load effect will not interfear with driver stages). The 2nd advantage that it will not need base current.
If I want to built this L-Mosfet as Driver and Bipolar as final stage, which is better, to put it as CFP or Darlington?
Transistor distortion also comes from the change in HFE, because of change in voltage or change in current. We can read this in cascode.pdf by Mr.Nelson Pass. Lavardin uses 1 single word (=thermal), while Mr.Pass uses 2words (voltage + current).
Will attach the transistor in a constant temperature heatsink (without keeping the voltage or current constant) will work Lavardin theory? Or the only solution is constant voltage and constant current?
Hi, Mike,
Yes, the ordinary configuration is like what you make. But somehow IGBT is "mosfet like" gate/base and "bipolar like" C-E/D-S. The advantage of mosfet in front that it will "block" the final stage from the previous driver stages (strange or heavy load effect will not interfear with driver stages). The 2nd advantage that it will not need base current.
If I want to built this L-Mosfet as Driver and Bipolar as final stage, which is better, to put it as CFP or Darlington?
lumanauw said:
hi lumanauw
Lavardin does not use only a single word but all 3 (thermal, current and voltage). But really these words are only used to explain the "memory" effect which is - at least for me - the real concept that can explain the "sound" resulting of the behavior of the amp characteristics modulated by the audio signal amplitude.
At first sight I do not know which would be better (increasing thermal inertia of transistor or keeping constant voltage and current). I believe that Lavardin suggests to limit all 3 as much as possible. To have constant voltage and current is less costly and takes less space than dissipator. I suppose that keeping constant power imply to have constant temperature at rhe same time.
In fact, the best way to convince ourself is always a listening test!
Having said that, it is not sure that all peoples prefer to hear the details which includes also the details of the bad recordings...The Lavardin web site discusses that matter.
Fab
I've tried the sound of cascodes (minimalizing voltage fluctuation) and sound of heavy biased transistor (minimalizing current fluctuation)----> in front stages, not output stages. They result in different way.
Cascoding a transistor gives better detailing. Putting heavy bias somehow -to me- results in "darker/less detailed" sound.
I never tried "constant dissipation", like Lavardin's differential.
In "peufeu" website, the author explained 4 steps. One step results in "darker" sound.
In theory, heavy biased transistor should give less distortion (which is current fluctuation originated). Why is heavy biased transistor gives "darker/less detailed" sound?
Cascoding a transistor gives better detailing. Putting heavy bias somehow -to me- results in "darker/less detailed" sound.
I never tried "constant dissipation", like Lavardin's differential.
In "peufeu" website, the author explained 4 steps. One step results in "darker" sound.
In theory, heavy biased transistor should give less distortion (which is current fluctuation originated). Why is heavy biased transistor gives "darker/less detailed" sound?
Hi lumanauw !
Yes, in sims cascodes are a good thing, especially in the VAS.
(haven't tried in real world yet, have no zeners in my box)
I think, if you use heavy biasing, the inputimpedance of these
transistors drop. If you don't compensate with some buffer, you
might loose highfreq-response.
But, i am no longer sure if this "ultradetailed" sound is correct.
Maybe in fact some overshooting or negative inputimpedances
for high freqs amplify these details/trebles, giving the typical hard
transistor sound.
But how do we know how it sounds "correct" and not only "nice" ?
I observed this "negative inputimpedance" in sims, it can be removed
with resistors and small caps, but i don't believe that sims are
accurate enough to find the right values. And i don't have an oscilloscope... 🙁
In sims i was able to linearize inputimpedance up to 20khz (+/- 1%),
giving a constant impedance of 22k over the audioband. (20hz-20khz)
Mike
Yes, in sims cascodes are a good thing, especially in the VAS.
(haven't tried in real world yet, have no zeners in my box)
I think, if you use heavy biasing, the inputimpedance of these
transistors drop. If you don't compensate with some buffer, you
might loose highfreq-response.
But, i am no longer sure if this "ultradetailed" sound is correct.
Maybe in fact some overshooting or negative inputimpedances
for high freqs amplify these details/trebles, giving the typical hard
transistor sound.
But how do we know how it sounds "correct" and not only "nice" ?
I observed this "negative inputimpedance" in sims, it can be removed
with resistors and small caps, but i don't believe that sims are
accurate enough to find the right values. And i don't have an oscilloscope... 🙁
In sims i was able to linearize inputimpedance up to 20khz (+/- 1%),
giving a constant impedance of 22k over the audioband. (20hz-20khz)
Mike
lumanauw said:
I have not tried heavy bias on my side but only constant dissipation with same typical bias in input and VAS stages. I was happy with the results.
Like MikeB says "...But how do we know how it sounds "correct" and not only "nice" ?..."
Lavardin also suggests to control the variation in DC offset too. Maybe class A power amp with its higher "steady" bias in output transistors (with no switching) suffer less of bias "variation" and even less temperature variation than class AB output stage?
From what I have read Lavardin's work was not only to get a patent but to really improve the sound of solid state amplifier since a lot of "audiophiles" prefered the "tube sound". Lavardin's discovery is that tube amp does not create "memory effect" (or at least not as solid state amps). They claim that their Lavardin amp has less "memory distorsion" than tube amp which in turn has less memory of "standard" solid state amp.
Fab
A question first :
What is the good biasing value for differential and VAS? Designs varies widely. One design use 0.5mA ccs for differential and 6mA for VAS. Other design use 50mA ccs for differential and 100mA for VAS.
I've tried those values, and they do sound different, but frankly I dont know which is the correct sounding one.
The majority of designs (bipolars) use 2.5-4ma ccs for differential and about 10mA for VAS. Is this the correct value?
But now you suggest this "input impedance" theory. Could you tell more about this?
What is the good biasing value for differential and VAS? Designs varies widely. One design use 0.5mA ccs for differential and 6mA for VAS. Other design use 50mA ccs for differential and 100mA for VAS.
I've tried those values, and they do sound different, but frankly I dont know which is the correct sounding one.
The majority of designs (bipolars) use 2.5-4ma ccs for differential and about 10mA for VAS. Is this the correct value?
If we use bipolars, it have HFE, meaning the more current in C-E it will need more current in base. At first I tought this "less detailed" sound in heavy biased bipolars are caused by this, that is if we use less bias in C-E, automaticly the base will be more sensitive to smaller signals (details). If this is true, then I think FET and Mosfet front stages wont suffer from this, like Passlab designs do not care about this and put very big bias for front end (because they use mosfets?)
But now you suggest this "input impedance" theory. Could you tell more about this?
I really want to understand about this "input impedance" and "negative input impedance". Could you tell more about this? An example schematic maybe? Also how to overcome this with your suggestion on "linearizing input impedance" with resistors and small caps?
Class A will fits nicely with both Lavardin theory or Mr.Pass paper. But making good sounding class B/AB is the real challange.
I never heard Lavardin or Lavardin-like solid state amp. You have built one. Is it sounds "not-like solid state"? Lavardin theory works in real SS amp?
hi !
The problem with HFE is, that it's the DC-transfer characteristic.
Typically this is the ratio Ie to Ib. This changes for higher freqs
due to internal capacities and speed of the device.
Normally you would expect that inputvoltage drops for higher freqs.
But circuits with feedback show some strange phenomena:
For higher freqs the inputvoltage increases...
I call this negative inputimpedance. I believe that it has to do with
phaseshifts.
Without feedback this effect is much lower, but still observable.
Of course this must have some effect on the sound, and it can make
an amp instable. (Like oscillating depending on the volumepot)
Most amplifiers compensate this with the inputfilter, a resistor (1-2k)
followed by a cap to ground. (I use 100pF)
The problem with FETs is, that their "HFE" is about infinite, but their
gates have a large inputcapacity. For expample the irf9540n: 1960pF !
Thats about 2nF... (irf610: 140pF,sk170: 30pF,bf245: 3pF)
You could call this reactive input.
I can show later a simple ckt, showing this effect.
In theory, this can create local resonances, increasing with higher
bandwidth. This might explain why an amp with high bandwidth
sounds brighter. Why should i need bandwidth far above 20khz ?
As long as the signal has only a constant delay, this shouldn't have
an impact on the sound. I don't understand it...
I once had an amp with bandwidth slightly above 20khz, and it
sounded boring, undetailed, lifeless. Just like my Yamaha.
Mike
The problem with HFE is, that it's the DC-transfer characteristic.
Typically this is the ratio Ie to Ib. This changes for higher freqs
due to internal capacities and speed of the device.
Normally you would expect that inputvoltage drops for higher freqs.
But circuits with feedback show some strange phenomena:
For higher freqs the inputvoltage increases...
I call this negative inputimpedance. I believe that it has to do with
phaseshifts.
Without feedback this effect is much lower, but still observable.
Of course this must have some effect on the sound, and it can make
an amp instable. (Like oscillating depending on the volumepot)
Most amplifiers compensate this with the inputfilter, a resistor (1-2k)
followed by a cap to ground. (I use 100pF)
The problem with FETs is, that their "HFE" is about infinite, but their
gates have a large inputcapacity. For expample the irf9540n: 1960pF !
Thats about 2nF... (irf610: 140pF,sk170: 30pF,bf245: 3pF)
You could call this reactive input.
I can show later a simple ckt, showing this effect.
In theory, this can create local resonances, increasing with higher
bandwidth. This might explain why an amp with high bandwidth
sounds brighter. Why should i need bandwidth far above 20khz ?
As long as the signal has only a constant delay, this shouldn't have
an impact on the sound. I don't understand it...
I once had an amp with bandwidth slightly above 20khz, and it
sounded boring, undetailed, lifeless. Just like my Yamaha.
Mike
Hi, Mike,
Thanks for the explenation. I can understand your concept of input impedances, but not very clearly.
About bandwith. Somewhere here, I read an expert said that a Hi-End power amp should have bandwith about 100Khz or more. Why? Because if we make that bandwith that far we can neglect the phase shift and magnitude distortions of signals up to 20khz. You can look at a simple speaker xover filter. Any L or C will introduce 45deg shift. That is big. You wont encountered noticeable shift in signals up to 20khz if you make bandwith about 100Khz or more. Not because we can hear sound up to 100khz. There's a reasonal cause for making such a high bandwith.
Off course making amp with low bandwith is easier, especially if we talk about oscilations.
Another question. You use JFET in your design. I begin to think about using FET or mosfet for front end, to advoid this HFE effect. In Borbely's site, it is stated how to use small JFETs. Mr.Borbely suggested that we use voltage cascode ---> to get steady voltage between D-S, and a source degeneration resistors (about 100ohm). I can understand the concept of using cascodes, to minimize Vd-s (that ends up with big gate capacitance fluctuation).
But I see in some design, they do not use source degeneration resistors. A design with JFET for differential just tied the sources together without any degeneration resistor.
Your design also tied the Sources of K170 together without degeneration resistors, and you also dont use voltage cascodes. (If you put cascode it will be Lavardin differential)
I've experimented with 2SK30 replacing MPSA06 in an audio power amp. The place is in input differential with CCS about 3mA, without any RE. I just replace it (D-C, S-E, G-B), without modifying anything else. The result-----> the sound is like having LPF in 5khz, all trebles are cut off. And the DC offset went crazy, originally about 15mV up to 170mV with 2SK30 (I admit I do not match the K30's). Why is this happens? Is treating JFETS without special care (put cascodes and source degeneration resistor) will make the input capacitance very big, so the original CCT perform LPF?
Thanks for the explenation. I can understand your concept of input impedances, but not very clearly.
About bandwith. Somewhere here, I read an expert said that a Hi-End power amp should have bandwith about 100Khz or more. Why? Because if we make that bandwith that far we can neglect the phase shift and magnitude distortions of signals up to 20khz. You can look at a simple speaker xover filter. Any L or C will introduce 45deg shift. That is big. You wont encountered noticeable shift in signals up to 20khz if you make bandwith about 100Khz or more. Not because we can hear sound up to 100khz. There's a reasonal cause for making such a high bandwith.
Off course making amp with low bandwith is easier, especially if we talk about oscilations.
Another question. You use JFET in your design. I begin to think about using FET or mosfet for front end, to advoid this HFE effect. In Borbely's site, it is stated how to use small JFETs. Mr.Borbely suggested that we use voltage cascode ---> to get steady voltage between D-S, and a source degeneration resistors (about 100ohm). I can understand the concept of using cascodes, to minimize Vd-s (that ends up with big gate capacitance fluctuation).
But I see in some design, they do not use source degeneration resistors. A design with JFET for differential just tied the sources together without any degeneration resistor.
Your design also tied the Sources of K170 together without degeneration resistors, and you also dont use voltage cascodes. (If you put cascode it will be Lavardin differential)
I've experimented with 2SK30 replacing MPSA06 in an audio power amp. The place is in input differential with CCS about 3mA, without any RE. I just replace it (D-C, S-E, G-B), without modifying anything else. The result-----> the sound is like having LPF in 5khz, all trebles are cut off. And the DC offset went crazy, originally about 15mV up to 170mV with 2SK30 (I admit I do not match the K30's). Why is this happens? Is treating JFETS without special care (put cascodes and source degeneration resistor) will make the input capacitance very big, so the original CCT perform LPF?
Hi !
Yes, i was thinking about the phaseshift. But if the phaseshift is
introduced by a constant delay, it shouldn't be audible !
Or is the delay not constant ? (Thus simply minimize it?)
Don't forget, if you sit 2 meters from the speaker, you alraedy
have a delay of 5.8ms. That's way above ~1us from an amp...
I didn't cascode the jFets in the diffamp, because voltagefluctuation
is far below 1v. (input ~0.4v, output of the diffamp ~2mv)
This is not much in compare to the voltagesupply.
If you use jFets in the diffamp, you need careful matching. The
variations between devices is BIG ! (>10%) This explains your
DC-offset. Did you look at the max Vds of the sk30 ? Typically
jFets have a low max Vds, this means eventually you overloaded
your sk30.
The sk170 is up to 40v, i use 24v at the moment.
most jFets goes up to only ~25v.
For higher supplyvoltages you must have cascodes, and according
to sims, jfets dont like high voltages. (leakagecurrent through
gate explodes,~1nA up to 1mA)
And jFets are different. For example for Vgs = 0v they are already
"open" (~2ma). And unlike mosfets, you can have a current through
the gate, but not in normal operation.
Except the facts of a negative Vgs-threshold and low max Vds, jFets
are not really difficult.
Mike
Yes, i was thinking about the phaseshift. But if the phaseshift is
introduced by a constant delay, it shouldn't be audible !
Or is the delay not constant ? (Thus simply minimize it?)
Don't forget, if you sit 2 meters from the speaker, you alraedy
have a delay of 5.8ms. That's way above ~1us from an amp...
I didn't cascode the jFets in the diffamp, because voltagefluctuation
is far below 1v. (input ~0.4v, output of the diffamp ~2mv)
This is not much in compare to the voltagesupply.
If you use jFets in the diffamp, you need careful matching. The
variations between devices is BIG ! (>10%) This explains your
DC-offset. Did you look at the max Vds of the sk30 ? Typically
jFets have a low max Vds, this means eventually you overloaded
your sk30.
The sk170 is up to 40v, i use 24v at the moment.
most jFets goes up to only ~25v.
For higher supplyvoltages you must have cascodes, and according
to sims, jfets dont like high voltages. (leakagecurrent through
gate explodes,~1nA up to 1mA)
And jFets are different. For example for Vgs = 0v they are already
"open" (~2ma). And unlike mosfets, you can have a current through
the gate, but not in normal operation.
Except the facts of a negative Vgs-threshold and low max Vds, jFets
are not really difficult.
Mike
Hmm, i will write a small program this evening, that simulates
a low bandwidth amp, and run this on a wav-file....
Maybe this answers some of my questions.
a low bandwidth amp, and run this on a wav-file....
Maybe this answers some of my questions.
lumanauw said:
From what I figure, for the input stage, the trick is maybe to use small current in input diff transistors only but use CFP arrangement where it is the other transistor that sees most of the current. As in peufeu experiment, you can have ccs of 7 ma and diff input transistor current of 0.7 ma and the rest of current 2.8 ma in the other transistor of the CFPs. This may resolve the results in "darker/less detailed" sound you have experimented.
For VAS current it depends of course of the selected diff input stage and next stage after VAS. Every stage must be considered to "balance" the bias of each stage. For me, it is always a matter of compromize unfortunately.
To avoid any confusion I have not built a Lavardin amp since I do not know the design in their real commercialized amps. I have only seen the Lavardin patent. I have adapted some of the Lavardin concepts in an existing amp. I do not pretend that it sounds like a Lavardin amp - in fact I have never heard such an amp myself but would be curious to. My goal is not to copy a Lavardin amp but rather use the low memory concept to design an amp. So far, I have modified the input and VAS stages but I want to apply the concept to the output stage also - but still using class AB.
Fab
I'm aware of their low Vds. In the amp the Vds is only 20V, so I think it is OK. But what makes the trebles dissappears?
FAB, is that the "darker/less detailed" sound is the correct one? I'm confused, which is the right one, the detailed sound, or the darker sound?
I also aimed at making good sounding class AB. VAS solution in Lavardin amp is putting 2 stacked cascodes. Is this differs if we put only 1 voltage cascode? Why it have to put 2 of cascodes?
hi lumanauw !
i dont understand why the trebles disappeared. Maybe some strange
design in this amp ?
I did some more sims, and detected some strange behaviour with
jfets, they tend to create overshooting like hell, no matter how big
cdoms are. Degenerationresistors reduced this effect, but then the
effect is simply slower. Very strange... I will do some more research,
otherwise i cancel jfets for me. I have a bjt-design with much more
OLgain, and it was easy to get completely rid of any overshooting.
Mike
i dont understand why the trebles disappeared. Maybe some strange
design in this amp ?
I did some more sims, and detected some strange behaviour with
jfets, they tend to create overshooting like hell, no matter how big
cdoms are. Degenerationresistors reduced this effect, but then the
effect is simply slower. Very strange... I will do some more research,
otherwise i cancel jfets for me. I have a bjt-design with much more
OLgain, and it was easy to get completely rid of any overshooting.
Mike
Lavardin
lumanauw
I do not pretend to know which sound is better. One thing I can say is that I prefer the detailed one because it is much enjoyable to hear all the stuff the musicians and arrangements have created. Otherwise, it is like they put so much efforts for nothing.
Lavardin patent does not describe what to do in the VAS from what I remember (they only correct the DC offset variation by using the VAS input instead of the diff. input stage). In peufeu, he has noticed that using 2 cascodes gives even better results in term of low memory. It is only a matter of deciding when to stop when trying to reduce the memory of a stage. On my side I used the darlington follower with cascode including magic resistor for the VAS. I also built a prototype with CFP-cascode VAS stage but I do not have verified so far the resulting phase margin of the whole amp.
Fab
lumanauw said:
lumanauw
I do not pretend to know which sound is better. One thing I can say is that I prefer the detailed one because it is much enjoyable to hear all the stuff the musicians and arrangements have created. Otherwise, it is like they put so much efforts for nothing.
Lavardin patent does not describe what to do in the VAS from what I remember (they only correct the DC offset variation by using the VAS input instead of the diff. input stage). In peufeu, he has noticed that using 2 cascodes gives even better results in term of low memory. It is only a matter of deciding when to stop when trying to reduce the memory of a stage. On my side I used the darlington follower with cascode including magic resistor for the VAS. I also built a prototype with CFP-cascode VAS stage but I do not have verified so far the resulting phase margin of the whole amp.
Fab
Hi, Mike,
This is the sch where I replace MPSA06 with K30. It is quite complicated, many cascodes--->to stabilize Vce. The main sch is that single differential, making push-pull VAS. The leg of left differential is driving upper VAS, while the right differential driving lower VAS with current mirror. The differential voltage is only +/-20V. Is the current mirror causing the K30 eliminating trebles? Or is it the input impedance for left and right differential performing LPF with K30 internal capacitance?
FAB,
This is the sch where I replace MPSA06 with K30. It is quite complicated, many cascodes--->to stabilize Vce. The main sch is that single differential, making push-pull VAS. The leg of left differential is driving upper VAS, while the right differential driving lower VAS with current mirror. The differential voltage is only +/-20V. Is the current mirror causing the K30 eliminating trebles? Or is it the input impedance for left and right differential performing LPF with K30 internal capacitance?
Dont give up on Jfets differential. I just about to begin one. Mr.John Curl succeeded in Jfet differential for years, with reputable power amp.
FAB,
That fits the majority of bipolar differential. They usually use <5mA ccs, resulting in detailed sound. While FET or mosfet (like passlab) doesn't care about this, they put very big bias for differential.
Attachments
hi lumanauw !
I am a little confused with your circuit...
You seem to have a "heavy" load on the output of the currentmirror,
but with "constant" voltage from the cascodes.
Isn't this against the idea of currentmirrors ?
I'll need further studying of your ckt...
Mike
I am a little confused with your circuit...
You seem to have a "heavy" load on the output of the currentmirror,
but with "constant" voltage from the cascodes.
Isn't this against the idea of currentmirrors ?
I'll need further studying of your ckt...
Mike
Hi, Mike,
It is quite simple cct, but the cascodes makes it difficult. Oh, I forgot to tell that in my experiment cct, I eliminated the 33ohm RE in differential (directly tied the emitors/sources together).
The left differential, via one cascode, is driving the upper VAS.
The right differential, via current mirror + one cascode is driving the lower VAS. The diode array (one in differential and one in cascodes) is just for protection. You can remove them to look at the basic sch clearly.
The current mirror here functioned as a shifter, signal from right differential, which is goes to +rail, is mirrored to opposite direction, to -rail. So it can drive lower VAS.
It is quite simple cct, but the cascodes makes it difficult. Oh, I forgot to tell that in my experiment cct, I eliminated the 33ohm RE in differential (directly tied the emitors/sources together).
The left differential, via one cascode, is driving the upper VAS.
The right differential, via current mirror + one cascode is driving the lower VAS. The diode array (one in differential and one in cascodes) is just for protection. You can remove them to look at the basic sch clearly.
The current mirror here functioned as a shifter, signal from right differential, which is goes to +rail, is mirrored to opposite direction, to -rail. So it can drive lower VAS.
lumanauw said:
lumanauw,
you misunderstood my point. I was suggesting to use high current CCS for diff input stage with low current in input bjts of the diff. Most of current goes into the current feedback bjt.
The jfet is more linear than the bjt. Jfet generally sustains lower current than mosfet. With jfets I have experienced higher DC drift in time than bjt (probably caused by thermal variation).
Fab
hi lumanauw !
I will have some closer looks to the ckt. Maybe you got the pinout
for the sk30 wrong ? Some jfets have g-d-s other d-g-s. I have no
idea why trebles should go.
btw, i played a little with your circuit, (the one you started the thread with)
I tried to create a NFB design. Its quite promisingly, the output from
VAS looks very proper. Nearly 2nd harmonic only, and still not very high.
I think your design is very good for NFB, what i like most is
the fact, that you have only 1 active transistor in input+vas ! (cfp = ~2)
(All other are ccs or else)
Because of the CFP i added, the 3rd harmonics have completely gone.
Bandwidth also more than enough, but the outputstage is the problem.
with a normal output, distortions are very high and dampingfactor
is a disaster. Looks like there's a need for a hawksford or something else.
Mike
I will have some closer looks to the ckt. Maybe you got the pinout
for the sk30 wrong ? Some jfets have g-d-s other d-g-s. I have no
idea why trebles should go.
btw, i played a little with your circuit, (the one you started the thread with)
I tried to create a NFB design. Its quite promisingly, the output from
VAS looks very proper. Nearly 2nd harmonic only, and still not very high.
I think your design is very good for NFB, what i like most is
the fact, that you have only 1 active transistor in input+vas ! (cfp = ~2)
(All other are ccs or else)
Because of the CFP i added, the 3rd harmonics have completely gone.
Bandwidth also more than enough, but the outputstage is the problem.
with a normal output, distortions are very high and dampingfactor
is a disaster. Looks like there's a need for a hawksford or something else.
Mike
Got it now. We can make big bias differential (if we use CFP), still the input bipolar is with small current (<1mA) and the rest will be held by CFP transistor.
Mr. Curl and Mr.Borbely likes to use servo. Maybe this is because the DC drift/temperature caused is too big like you observed? Inspite of its DC offset weakness, they always use Jfet for input differential. Maybe the sound is too good so they dont want to change to bipolar input differential?
I think I made no mistake. But I will re-experiment it again.
My design has already NFB, with 100K feedback and 1K input R, hasn't it?
Bad output stage, eh?😀 How about parrareling the output device, will it help?
or do you have simpler solution for the problem you found, but please not so difficult as putting Hawksford error correction.
Damping factor is a dissaster. What is the sound of audio amp with dissaster damping factor? Bad bass?
- Status
- Not open for further replies.