The low input impedance will always give better treble (by theory). The opposite is not always true. Rough statistics (tendency/likeliness) is the higher the MOSFET power, the higher input impedance, the higher maximum current, the better bass performance.
Besides, if you got much treble, you would think that you don't have the bass (while the frequency is in fact still there)
Besides, if you got much treble, you would think that you don't have the bass (while the frequency is in fact still there)
This is an interesting and true observation. Different Mosfets often measure about as well in the bottom/mid/top frequencies but still have different sounds. The larger ones tend to have better/more bottom end due to their higher transconductance and the slight de-emphasis of the top and the reverse is true for the smaller ones. This effect also applies to paralleling Mosfets, as big Mosfets are simply the same sort of thing but with more land area.
Bass and treble
Any calculations ( formula) to see this effect ?
While a treble lift can make the sound appear to have less apparent bass , will this cause such a large variation in treble to make it effective in this case? The devices will be used in amplifiers with negative feedback in most cases ,which widens the frequency response.
While individual devices, in say Class A amps with no feedback, these factors may become audible between different types of devices, is it still applicable to systems with NFB? If so are there any commercial examples we can hear or is it relegated to the design labs.
Thanks.
Any calculations ( formula) to see this effect ?
While a treble lift can make the sound appear to have less apparent bass , will this cause such a large variation in treble to make it effective in this case? The devices will be used in amplifiers with negative feedback in most cases ,which widens the frequency response.
While individual devices, in say Class A amps with no feedback, these factors may become audible between different types of devices, is it still applicable to systems with NFB? If so are there any commercial examples we can hear or is it relegated to the design labs.
Thanks.
Re: Bass and treble
No calculations from my end. You can see subtle effect in measurement such that you wouldn't expect a really audible difference, but then you hear it.
Commercial examples? Aleph 3, 5, 2, and 1.2. they all have virtually identical circuits except power supply and output stage, and they all have a different signature, and it largely follows this effect.
I don't worry all that much in any case. Like a batch of wine, it is what it is, and you can enjoy that and then swap something else in and enjoy that too.e96mlo said:
ashok said:
No calculations from my end. You can see subtle effect in measurement such that you wouldn't expect a really audible difference, but then you hear it.
Commercial examples? Aleph 3, 5, 2, and 1.2. they all have virtually identical circuits except power supply and output stage, and they all have a different signature, and it largely follows this effect.
Ashok,
While I did talking about formulas, I guest you and Mr. Pass were more concerned with the calculation result. Well, (right or wrong) here is my thought regarding both formulas and calculation result:
Even if the output impedance of the MOSFET driving stage is low, to drive the capacitance, or to charge the capacitor, current is required, right? How much is this idle current required from the driving stage?
Current = Slew Rate * Capacitance
Slew Rate = (2 * 3.14 * F * Vpeak ) / 1000000
F = I / (C * V * N)
Noted = We need lower capacitance to “slew” higher frequency, especially if current is low as in MOSFET pre-amplifier.
Again, capacitance (C) is “evident” in the Gate of any MOSFET (Anybody knows how to measure it in a given situation as in BOZ?). Then we must be able to “virtually” construct an AC equivalent circuit as seen by the capacitance to calculate the R thevenin (R).
Half-power frequency = 1/(2*3.14*C*R), is the frequency where significant loss occurs.
I don’t think that measuring low capacitances is accurate with standard measurement tools. How about measuring input capacitance of a 3-pole Source grounded MOSFET? Will it be easier or harder? Well, the specs is simple: Ciss=Cgd+Cgs, and in a condition (voltage, frequency, etc) that does not apply in the real circuit. Also, manufacturers tend to give “optimistic” values to customers (I guess there is not many situation where a high input capacitance is preferable, if any).
Out there, there is always a strong debate regarding the use of “hi-end” cables in audio. If you bring the issue to an algebra professor, I think he will come with a conclusion that the capacitance (or impedance) or any available metrics will not give much effect in audio frequency. Also, in the design of an amplifier’s frequency respond (or a low pass filter) I noticed as if the designer found some “advantages” to make the respond flat through-out a wide band which is not in audio frequency. May be the formulas are not applicable enough, or the designer just don’t know how to use the formula properly, or may be it’s just the time for “golden-eared” audiophiles to take action
(Golden eared or not, unfortunately I can clearly hear the diference! It's not like the amplifier having a usual ealier roll-off. Rather, it's like the graph is truncated in an audible high frequency )
While I did talking about formulas, I guest you and Mr. Pass were more concerned with the calculation result. Well, (right or wrong) here is my thought regarding both formulas and calculation result:
Even if the output impedance of the MOSFET driving stage is low, to drive the capacitance, or to charge the capacitor, current is required, right? How much is this idle current required from the driving stage?
Current = Slew Rate * Capacitance
Slew Rate = (2 * 3.14 * F * Vpeak ) / 1000000
F = I / (C * V * N)
Noted = We need lower capacitance to “slew” higher frequency, especially if current is low as in MOSFET pre-amplifier.
Again, capacitance (C) is “evident” in the Gate of any MOSFET (Anybody knows how to measure it in a given situation as in BOZ?). Then we must be able to “virtually” construct an AC equivalent circuit as seen by the capacitance to calculate the R thevenin (R).
Half-power frequency = 1/(2*3.14*C*R), is the frequency where significant loss occurs.
I don’t think that measuring low capacitances is accurate with standard measurement tools. How about measuring input capacitance of a 3-pole Source grounded MOSFET? Will it be easier or harder? Well, the specs is simple: Ciss=Cgd+Cgs, and in a condition (voltage, frequency, etc) that does not apply in the real circuit. Also, manufacturers tend to give “optimistic” values to customers (I guess there is not many situation where a high input capacitance is preferable, if any).
Out there, there is always a strong debate regarding the use of “hi-end” cables in audio. If you bring the issue to an algebra professor, I think he will come with a conclusion that the capacitance (or impedance) or any available metrics will not give much effect in audio frequency. Also, in the design of an amplifier’s frequency respond (or a low pass filter) I noticed as if the designer found some “advantages” to make the respond flat through-out a wide band which is not in audio frequency. May be the formulas are not applicable enough, or the designer just don’t know how to use the formula properly, or may be it’s just the time for “golden-eared” audiophiles to take action
(Golden eared or not, unfortunately I can clearly hear the diference! It's not like the amplifier having a usual ealier roll-off. Rather, it's like the graph is truncated in an audible high frequency )
A lot of this depends on which capacitor you are charging,
essentially GS or GD. GS is the large capacitance, but the
voltage change across it reflects only the Vgs change, which
is low and inversely proportional to the transconductance,
which in turn is proportional to the capacitance....
Cdg is a lower value, but in many circuits will see a larger
voltage change.
To measure the effective capacitance in a given configuration,
you could place resistance in series with the Gate and
measure the response rolloff at higher frequencies versus
the Gate current. If one is constant, then when you get
a 3 dB change on the other measurement, you have reached
the 1/RC = 6.3 * Freq, and this gives you C.
essentially GS or GD. GS is the large capacitance, but the
voltage change across it reflects only the Vgs change, which
is low and inversely proportional to the transconductance,
which in turn is proportional to the capacitance....
Cdg is a lower value, but in many circuits will see a larger
voltage change.
To measure the effective capacitance in a given configuration,
you could place resistance in series with the Gate and
measure the response rolloff at higher frequencies versus
the Gate current. If one is constant, then when you get
a 3 dB change on the other measurement, you have reached
the 1/RC = 6.3 * Freq, and this gives you C.
So OK this may sound like a strange suggestion, What if you were to mix IRFP140's and 150's or 240'?
In an Aelph 2 where you have 12 Devices you could have 2 of each on each side. Matching them could be tricky
Would thier individual character meld into one, or would the poor thing go mad like some Frankenstien abomination.
Hummm .....too much Shiraz tonight.
Anthony
In an Aelph 2 where you have 12 Devices you could have 2 of each on each side. Matching them could be tricky
Would thier individual character meld into one, or would the poor thing go mad like some Frankenstien abomination.
Hummm .....too much Shiraz tonight.
Anthony
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.