I don't care for them because I am suspicious of the RF they put out. Of course I have a CD player also....Bernhard said:
Using switching supplies, keep in mind that unlike most linear supplies, they have no headroom above spec. If your amplifier will draw a peak current greater than the spec, it will run out of juice, so figure on specifying current as follows:
R = minimum load impedance, ohms
P = rated power into minimum load impedance
I = power supply current rating, amps
I = SQRT(2 * P / R)
This leaves no headroom, so think about making I a bit larger.
Hello Mr. Pass,
do you have a problem in general with using 12 volts rails for the amp or just with the voltage coming from a switching PS?
How much 'dirtier' does the power get in going from an aleph3 with 25 v rails to a zen 4 with a 20V swing or a low power SOZ?
I know the future holds bigger and better versions of the AX for me, but I was sort of happy with my AX prototype being driven by 12.5-0-12.5, now you got me all paranoid about this!
do you have a problem in general with using 12 volts rails for the amp or just with the voltage coming from a switching PS?
How much 'dirtier' does the power get in going from an aleph3 with 25 v rails to a zen 4 with a 20V swing or a low power SOZ?
I know the future holds bigger and better versions of the AX for me, but I was sort of happy with my AX prototype being driven by 12.5-0-12.5, now you got me all paranoid about this!
I dont know if it has been covered before, but what about connecting the PSU caps across +/- and not +/ground/- ?
that would quadruple the stored energy and the symetrical nature of the X should not mind?
/ Micke
that would quadruple the stored energy and the symetrical nature of the X should not mind?
/ Micke
hifi,
The problem was addressed, maybe, in the X100 thread but it was a long time ago.
Another question for Mr. Pass I guess, you can probably play it safe and have both, a standard ps with an additional cap from + to -.
The problem was addressed, maybe, in the X100 thread but it was a long time ago.
Another question for Mr. Pass I guess, you can probably play it safe and have both, a standard ps with an additional cap from + to -.
Thanks for the formula 🙂
May be with feedthrough caps and other filtering and an additional big cap RF should be no problem.
Anyway, I have a plasma tweeter, what shall I worry about RF from a switching power supply
Another idea to reduce wasted ClassA idle watts. 
If a +/- 25V Amp will still work with +/- 12.5V, just with reduced output power, why not use a regulted PSU and couple that with the volume control ?
If a ClassA amp draws constant current...
If a +/- 25V Amp will still work with +/- 12.5V, just with reduced output power, why not use a regulted PSU and couple that with the volume control ?
If a ClassA amp draws constant current...
grataku said:do you have a problem in general with using 12 volts rails for the amp or just with the voltage coming from a switching PS?
How much 'dirtier' does the power get in going from an aleph3 with 25 v rails to a zen 4 with a 20V swing or a low power SOZ?
I know the future holds bigger and better versions of the AX for me, but I was sort of happy with my AX prototype being driven by 12.5-0-12.5, now you got me all paranoid about this!
12 volts is just a bit low. Don't get excited, except for your next amp, which will no doubt have more power....
You still need something +/g/-, but given that you can parallelhifi said:
additional caps + to - just fine.
The Aleph X is pretty sensitive to supply voltage with respect to absolute DC offset. Every time you adjust the the rails, you'll also need to readjust the absolute offset: And don't ask me about servos to fix that problem.Bernhard said:
😎
Servos just aren't elegant. I much prefer to make the circuit
behave itself in the first place. Sometimes that's not practical.
Servos are often justified by the benefits of removal of a cap
from the main feedback loop, but we are just substituting it
with another (ok, probably better) cap and an active circuit.
I personally think that there are bigger devils out there than the
lowly, much maligned capacitor. Any active gain device or
transformer is a much bigger offender, and deserves more
attention than wire or capacitors.
behave itself in the first place. Sometimes that's not practical.
Servos are often justified by the benefits of removal of a cap
from the main feedback loop, but we are just substituting it
with another (ok, probably better) cap and an active circuit.
I personally think that there are bigger devils out there than the
lowly, much maligned capacitor. Any active gain device or
transformer is a much bigger offender, and deserves more
attention than wire or capacitors.
Nelson Pass
Steve McCormack agrees with you. In the mods he performs on the old DNA series, the very first things he removes is the servo from his circuit. I'd like to ask a question about matching the P-channel devices in the Aleph-X. I am of the opinion that the speed or beta characteristic of the input FETs is much more important than the mV reading we get from an applied and constant voltage we use in the test circuit. The output FETs will eventually settle so matching method in that case is viable but I feel uneasy about the same methodology for the P-FETs. How can we reliably test for beta? Am I way off track?
Steve McCormack agrees with you. In the mods he performs on the old DNA series, the very first things he removes is the servo from his circuit. I'd like to ask a question about matching the P-channel devices in the Aleph-X. I am of the opinion that the speed or beta characteristic of the input FETs is much more important than the mV reading we get from an applied and constant voltage we use in the test circuit. The output FETs will eventually settle so matching method in that case is viable but I feel uneasy about the same methodology for the P-FETs. How can we reliably test for beta? Am I way off track?
Since Fets, Mosfet, and tubes do not have current gain as such, transconductance is the term you want.
Matching Vgs for the Fets and Mosfets is pretty effective in getting overall matches if the devices come from the same die, which I take to mean the lot code.
On more than one occasion I have found it helpful to measure the transconductance of Mosfets as part of the matching procedure, and in the case of many Mosfets it is equally important to measure transconductance versus frequency, as this varies for a lot of the chips we use. The IRF610 and its relatives, for example, go through about a 6 dB transition in transconductance through the audio midband, and choosing your Mosfets based on variations in this figure can make a performance difference.
(Bet that transition doesn't show up in your Spice model, either.)
I have a nice little circuit for this, and I'll dig it up and post it.
Better yet, some time ago we began matching Mosfets right in the circuit, plugging them into sockets and measuring various performance characteristics. This is a bit tedious, but gets terrific results.
Matching Vgs for the Fets and Mosfets is pretty effective in getting overall matches if the devices come from the same die, which I take to mean the lot code.
On more than one occasion I have found it helpful to measure the transconductance of Mosfets as part of the matching procedure, and in the case of many Mosfets it is equally important to measure transconductance versus frequency, as this varies for a lot of the chips we use. The IRF610 and its relatives, for example, go through about a 6 dB transition in transconductance through the audio midband, and choosing your Mosfets based on variations in this figure can make a performance difference.
(Bet that transition doesn't show up in your Spice model, either.)
I have a nice little circuit for this, and I'll dig it up and post it.
Better yet, some time ago we began matching Mosfets right in the circuit, plugging them into sockets and measuring various performance characteristics. This is a bit tedious, but gets terrific results.
Nelson Pass
Well, the matching circuit available on the A75 project does a good job of transconductance matching as I understand it but what I was trying to refer to was the "transconductance profile" if such a term exists. Will I need a frequency generator to plot the variations you described above? The "beta" factor I was trying to describe has more to do with the reactive speed of the silicon in the mosfet or the latency, if you will. Should I assume that the same lot code will result in identical latency on the chips? The current source mosfets will eventually settle to a value (Vgs) but the input signal mosfets have to contend with a constant voltage adjustment from the input signal variations. It seemed to me that the matching technique used for the output devices would not be suitable for the inputs but I couldn't think of any other way than building the whole circuit and swapping till satisfied. Am I to understand that even the output FETs, are also subject to these variations? It would be great if you had another way to do it, I'm all ears!
Yeah!!! 🙂 🙂 🙂
My prototype Aleph-X is working. It is the Greg version with the PCB v1.0 bought from hifizen. Everything looks ok. I have about 0.02V of DC offset between out- en out+ and about 1.2V absolute DC offset.
I use Petter his V2=1K, R24=296 and R26=568 to control absolute DC offset but I do not have enough control over the offset. I used these as I could not get a V2=200 from my shop so I can try the default setup. I will hopefully get these next week.
My shaky power supply starts to sag 😎 at about +/- 10V of output. But as this is only a prototype composed of two 100VA trafo's put together to get to a +/- 15V. When everything is working I will get started to work on a dual mono 140W version (7 amps of bias and 26V).
Tomorrow I will continue...
Edwin
My prototype Aleph-X is working. It is the Greg version with the PCB v1.0 bought from hifizen. Everything looks ok. I have about 0.02V of DC offset between out- en out+ and about 1.2V absolute DC offset.
I use Petter his V2=1K, R24=296 and R26=568 to control absolute DC offset but I do not have enough control over the offset. I used these as I could not get a V2=200 from my shop so I can try the default setup. I will hopefully get these next week.
My shaky power supply starts to sag 😎 at about +/- 10V of output. But as this is only a prototype composed of two 100VA trafo's put together to get to a +/- 15V. When everything is working I will get started to work on a dual mono 140W version (7 amps of bias and 26V).
Tomorrow I will continue...
Edwin
I just found an old 250 ohm pot, so I changed to the Grey settings... and... I could get rid of the absolute DC offset... The pot is not the most stable version available, but I can get it to about 0.1V or so (I have a range from -2 to +6 or so!)...
So it seems Grey's values are the golden ones!
Edwin
P.S. I use the default 10K and 4k7 for the other magic resistors!
golden values
For the most part but not always. It really depends on what you CCS you use and what sound you want to get. For example, many people have told me that 68K in the input gives way too forward treble.
For the most part but not always. It really depends on what you CCS you use and what sound you want to get. For example, many people have told me that 68K in the input gives way too forward treble.
In previous posts, Nelson said that +12 0 -12 rails were not quite enough for the Aleph-X
What would minimum rails be??
What would minimum rails be??
It's not that it won't work, just that higher voltages work much better, say +/- 20 + volts. Myself, I would go with 24 or 25 volts.