Many have eluded or hinted at this among the many postings of this thread, but studying the operations of this mu follower topology has really helped me understand some of the happenings of the Aleph CS, ie current feedback, negative AC impedance. I am not saying i get it yet, but I am think i am getting there. I believe it has been stated already, but the biggest problem with this topology with these particular fets is the gain. Since they do not have a significant (ra = plate resistance = output impedance), no voltage divider is formed reducing this gain to usable levels. All my attempts set up as suggested has resulted in a output that is clipping due to, what i think is the DC bias issue that has been mentioned. I have been unsuccessful in my attempts with self-biased setup to throw away enough of this gain to gain stability.
Last edited:
The requirement for a negative bias is not always an advantage.
Nevertheless, it often comes with the most interesting parts, so there it is.
😎
Nevertheless, it often comes with the most interesting parts, so there it is.
😎
Indeed. Hopefully I will figure it out enough to build something soon. Meanwhile, I am going back to chew on ZV2 some more.
Not particularly. I have been trying all sorts of different parts there, and
certainly they all work. As development proceeds, one part may emerge
as the best, but that work is not completed. Keep in mind that there are
only three copies, and they are all prototypes.
😎
certainly they all work. As development proceeds, one part may emerge
as the best, but that work is not completed. Keep in mind that there are
only three copies, and they are all prototypes.
😎
Nelson,
How much variation did you see in your r085 fets. At Vds of 22V and 2A I have a Vgs of 2.90. This is no where near the datasheet. If i remember correctly, you don't let the Jfets heat up when you measure them, correct? I am going to pull together a version of the F2 with dual 22V rails, using the bottom rail to set drain potential as you suggested. Hopefully this will give me an idea of operating points and sound. I will not get my diodes in time to try you negative voltage generator idea. Considered wring two trafos together to raise single rail voltage high enough to try 100V De Lite, but I am not real confident in the idea and worry about proper current sharing.
How much variation did you see in your r085 fets. At Vds of 22V and 2A I have a Vgs of 2.90. This is no where near the datasheet. If i remember correctly, you don't let the Jfets heat up when you measure them, correct? I am going to pull together a version of the F2 with dual 22V rails, using the bottom rail to set drain potential as you suggested. Hopefully this will give me an idea of operating points and sound. I will not get my diodes in time to try you negative voltage generator idea. Considered wring two trafos together to raise single rail voltage high enough to try 100V De Lite, but I am not real confident in the idea and worry about proper current sharing.
Last edited:
buzzforb said:
Compared to what? With a typical Pass design SE FET amplifier, the FET is loaded by the 8ohm speaker in parallel with the drain load, so if the drain load is less than an infinite impedance, the FET is seeing <8ohm. With a 1:1 transformer, the FET will see an 8ohm load with an 8ohm speaker....
Roscoe
Last edited:
Look at figure 2 in ZV3. Regardless whether you use it as a regulated supply, or lose the zeners and use it as a capacitance multiplier, if you make the time constant of R1C1 large, you'll get a slow turn-on. In my DeLite amp, I'm using 5.11K/560uF and there's NO turn-on thump.... Takes >5sec before the bulbs even start to glow.
Roscoe
You have probably not noticed that the data sheet only goes to a few volts
Vds. At 20V, 2A, -2.9 volts looks fairly typical.
If I am simply matching parts, I usually do it consistently quickly, with about
2 seconds delay. If I am measuring to get absolute values, I will take the
time and trouble to put it at the current and voltage of interest and let it
warm up.
😎
If you can afford to lose 4 volts or so, that works very well.
If you are using depletion mode parts (Common-Source mode) with a
resistive or CCS type load and want a slow turn on, bias the Gate to the
Fet slowly, and it will start up in full conduction (ie low voltage) and slowly
rise.
😎
Speaking of which, I notice you generally use FETs rather than bipolars (BOZ a notable exception) for regulator pass transistors/capacitance multipliers. Is there a performance advantage to FETs in this application, or is this another case of using what's on the shelf?
Roscoe
Mosfets don't require any control current to speak of, and so you can drive
the Gates from a high resistance decoupled by fairly small capacitance.
Also, they typically have enough Vgs drop that they will regulate over a
3 volt noise/fluctuation if driven just by an RC.
That and I have buckets of them...
😎
the Gates from a high resistance decoupled by fairly small capacitance.
Also, they typically have enough Vgs drop that they will regulate over a
3 volt noise/fluctuation if driven just by an RC.
That and I have buckets of them...
😎
Yeah, it had occurred to me that the bipolars might not be good in a capacitance multiplier unless the supply had very low ripple anyway. As a regulator, it does give one a little more flexibility on regulated voltage to have both FETs and bipolars if you only have limited choices of zeners to play with (I have a boatload of 12V zeners, but not mush else,) just need to make sure the unregulated voltage is enough higher than the desired regulated voltage.
Roscoe
Roscoe
I like to use a nice big capacitor after the Mosfet to help filter out any
artifacts of the Mosfet. That way I can feel assured that the character is
mostly passive.
😎
artifacts of the Mosfet. That way I can feel assured that the character is
mostly passive.
😎
Give me a few days to finish tweaking it... Depletion mode MOSFETs, source boased with no input cap, adjustment via the current regulator, which is pretty tweaky at this point....
Built into a re-tasked Adcom GFA-555 II chassis ;-)
Roscoe
Built into a re-tasked Adcom GFA-555 II chassis ;-)
Roscoe
For lowish voltages, capacitace multipliers with power Darlingtons work quite well - so do mosfets, but you get more voltage drop. I use mosfets at higher voltages because of better SOA and more unambiguopus voltage ratings. With bipolars, is it VCEO. VCER, VCEX, VCEV (???), all the ratings dreamed up by marketing cheese-parers to make a high voltage transistor look splendid on paper until you actually try to use it in a real circuit. For Darlingtons in this application, VCEO is probably the voltage rating most applicable.
A qualified yes - I've been using either surplus Lambda Darlingtons or 2N6283 series devices, also TIP141/TIP147. If the TIPs work, I guess the old Motorola Darlingtons would do ok, too.I really need to put a current probe on the pass devices of my cap multipliers to see if they behave themselves during startup or whether they get subjected to a blast of current - yet another thing to do...
- Status
- Not open for further replies.