| Kashmire |
Choke loads have been informally discussed before:
http://www.diyaudio.com/forums/show...7393&highlight=
Time to look a little closer at the driving factors of why we build stuff the way we do.
Be nice to the tube guy, now. Looking at the Zen design (and the Aleph design), a current source is used to force an output when the amplification MOSFET modulates. In the tube world, we call this "parafeed".
In parafeed amps, a choke is used as the current source. The DC bias current shoots through the choke and into the tube. When an AC signal appears at the tube, the choke resists the AC signal, and forces the signal to go elsewhere - the output.
If the AC signal is low enough in frequency, it appears like slowly changing DC to the choke. The choke doesn't try to fight it, so nothing appears at the output.
This theory can be applied to Zen amps (Figure below). The Pass design uses a current source that burns 50% of the voltage as heat in Figure (a). In Figure (b), the choke doesn't burn any voltage, except a little due to DCR.
Now, tube amps usually use about 100mA, and chokes are relatively easy to make for lower currents. The Zen amps use an order of magnitude more current, which means the chokes have to be that much larger. Pass Labs probably selected not to use chokes because it may be financially and commercially infeasible.
However, for the DIYer, chokes may be acceptable. The tradeoff is as follows: higher construction costs of the very large choke will result in much less power dissipation, less heat sinks, and cheaper operating costs. Those who are thinking about water-cooling their amps may find relief. Those who are concerned about ecology may also find solace, not wasting as much energy as heat.
Is there sonic differences? A tube guy would think not. However, those die-hard may find Figure (c) desirable: not only does the active current source exist, but a passive source also exists. This does not save any heat/energy dissipation problems, but it may increase the fidelity of the design. |
|
|
| tiroth |
Well, the impedance of an active current source is generally so high as to swamp any effect of the impedance of the choke. I'm not sure I see the point...
Since all of the current is passing through either way it seems to me that you either bite the bullet and use huge iron or just go active. |
|
|
| Kashmire |
Let me make a correction: Figure (c) may be considered a natural progression from (a) and (b). If each of the things is good independently, would they be even better combined? This was more of a rhetorical question to spawn discussion.
What is more germane to the discussion is why DIY hobbyists have chosen (a) more often than (b).
If chokes may not be financially feasible for commercial products (such as Pass Labs), are chokes feasible for DIY projects? I believe that chokes are within the scope of the average DIY hobbyist, at least on this forum.
Pass has tried chokes before (refer to previous thread) and had good results. Why do the Pass Labs amps not all have chokes? In general, some things may be more feasible for commercial operations – such as mass produced circuit boards or anodized aluminum faceplates. However, other things may not be feasible for commercial operations. High-current chokes may be one of those things. That may be the primary value of Pass’s patents: giving an amplifier choke-like performance without the cost of the choke.
As a DIY hobbyist, high-current chokes may be feasible. Do the performance advantages warrant exploring the option more often?
This may be implying that Pass’s invention is a “band-aid” to get around using a choke. I don’t endorse that theory. I endorse the theory that Pass’s invention was a way to bring his vision of high-quality audio to commercialization. Otherwise, it may have not been possible. |
|
|
| Magura |
Forgive my lack of knowledge, but wouldnt those chokes have to be size x-large to be useful for bias in a SOZ or is there something ive gotten wrong?
I made a set of 4 pcs 2mH chokes for the psu for my SOZ project (air gap) and they are simply huge (600W a piece)
Magura |
|
|
| Nelson Pass |
This subject has been addressed before (in this forum, by me,
don't ask where) and I repeat:
Chokes (especially air core) are great loads for Alephs, and
I have built such with a ~ 1 H load using a full roll of MWS
magnet wire. The bias is a little tricky (left for you to either
work out or wait for an article) but the performance is excellent,
with an symptotic approach to 50% efficiency and just about
the specs you get from an Aleph current sourced Zen.
The roll of 1000 ft of MWS wire ( I forget the gauge - probably
16 or 18) doesn't cost much <$100 but it does take up a little
space on top of the box.
The bias is a little more sensitive, but nothing you can't trim -
I recommend a meter on the front panel with a knob.
:wiz:
Oh, I forgot to mention, when you do this, you tend to do it
at 1/2 supply rail with twice the current. |
|
|
| Fuling |
How ´bout some choke loaded IT-coupled balanced SE tube/mosfet hybrid action???:D
I have all the iron except the power transformer, which should be easy to find used or surplus. Shall I??? |
|
|
| Magura |
I find your idea interesting.
I think i once saw an inductor calculator online...in that case it should be no problem to make the 50mH inductors....though they are going to become huge.
Are you sure 50mH is enough?
Magura |
|
|
| Fuling |
I *think* 50mH will be enough, at 20Hz they have a reactance of 6,28 ohm and since the circuit is bridged each half should "see" a 4 ohm load, correct me if I´m wrong. I guess the higher inductance the better, but this should work.
Experience shows that 100mH is enough for an unbalanced SE circuit loaded with 8 ohms.
The chokes I´m planning to use are 25mH @ 3A, double C-cores.
Have a look at page 11 in the thread Circlotron linked to if you wanna see a pic of my earlier attempt to use them in an amp.
Just a thought: In a balanced circuit it should be possible to use one centertapped choke with just a small airgap instead of two separate chokes gapped for 3A.
Though, I guess that would make it a PP amp which could be biased in class A, AB or B. Not funny anymore... |
|
|
| Magura |
There it was...
http://www.colomar.com/Shavano/inductor_info.html
The inductor calculator that is.
Looking at a 50mH air gap you can achieve 8.81-1.13 ohms, but wasnt the point to bias the mosfets without the low efficiency??
If we see the same resistance as with the resistors, but have added some monster inductors....whats the benefit then??
Magura |
|
|
| Fuling |
Uhm, you might want to use some iron cores for the chokes, otherwise they will be huge and the DC resistance will be way too high.
Again, have a look in the thread Circlotron posted, there he shows how to make really high inductance chokes out of old microwave oven transformers. |
|
|
| Magura |
True...no problem to make some 50mH inductors of low resistance.
If needed it would maybe be even easier to just make the E and I cores yourself, in order to obtain a bunch of matched inductors.
All of a sudden a SOZ could be quite efficient :)
Magura |
|
|
| Fuling |
I don´t think DIY core laminates will be easy, but has anyone ever tried actually?
Yes, the maximum theoretical efficiensy of a class A SE choke loaded amp is 50%, not that bad. I reached about 40% which isn´t bad either I think. |
|
|
| Magura |
| quote: | Originally posted by Fuling
I don´t think DIY core laminates will be easy, but has anyone ever tried actually?
Yes, the maximum theoretical efficiensy of a class A SE choke loaded amp is 50%, not that bad. I reached about 40% which isn´t bad either I think. |
Ive machined cores for trafos before, no big deal.....epoxy does wonders :)
Sounds rather appealing...especially if you think of the original SOZ power efficiency.
Magura |
|
|
| Fuling |
SOZ has like 4% efficiency or something, right?
If i had the necessary heatsinks lying around I´d love to build one...:) |
|
|
| Magura |
true....how to calculate the efficiency of a SOZ with chokes??
I have the sinking laying around....actually im in the middle of making a SOZ, but the choke thing seems very attractive.
Magura |
|
|
| Fuling |
| quote: | | how to calculate the efficiency of a SOZ with chokes?? |
Uhm, I don´t know. I guess those "tail" resistors (from sources to negative rail) dissipates alot of heat, but maybe they can be replaced with a third choke, acting as a constant current source?
I don´t know, I´m just thinking loud here.
BTW with chokes instead of resistors in the SOZ the gain probably goes way up so some kind of negative feedback might be necessary. Not attractive.
Thinking about cores, as I see you live in Denmark maybe you could have a look at www.svebry.se and check out their EI transformer kits. Much easier than ripping old power transformers apart. |
|
|
| Magura |
| quote: | Originally posted by Fuling
Uhm, I don´t know. I guess those "tail" resistors (from sources to negative rail) dissipates alot of heat, but maybe they can be replaced with a third choke, acting as a constant current source?
I don´t know, I´m just thinking loud here.
BTW with chokes instead of resistors in the SOZ the gain probably goes way up so some kind of negative feedback might be necessary. Not attractive.
. |
Sure not attractive...so this is most likely the end of chokes for SOZ....or what??
Great link...postage could be mean though :(
Magura |
|
|
| Fuling |
I suspect that the easiest way to make a mosfet work good with at choke load is to use it as a source follower and use a preamp for voltage gain.
Applying feedback to a choke loaded Zen-like amp seems to be a ***** unless the rail voltage is well regulated, the DC resistance in the choke is too low to make the feedback loop work DC-wise.
Yeah, postage...:bawling:
Maybe it would be cheaper to pick them up yourself??:D
BTW, you can also get custom wound transformers and chokes from Svebry and they don´t charge much for the work. |
|
|
| Fuling |
| Wow, I got censored:cannotbe: :cannotbe: |
|
|
| Magura |
| quote: | Originally posted by Fuling
Yeah, postage...:bawling:
Maybe it would be cheaper to pick them up yourself??:D
BTW, you can also get custom wound transformers and chokes from Svebry and they don´t charge much for the work. |
Well, the postage prob seems to have solved itself....since if a SOZ cant be made using chokes, i wont need any....
If i ever would need a trafo from sweeden, the only way to get it here cheaper than it could be made here....would be to swim to sweeden to pick it up.....we have a few comps in DK that winds custom trafos as well, without charging an arm and a leg (if you know the right people that is) :)
Magura |
|
|
| dhaen |
| quote: | Originally posted by Fuling
Wow, I got censored:cannotbe: :cannotbe: |
Yes, I lost a female dog on a post I made :xeye: Seems to be automatic.
BTW: Interesting thread :) |
|
|
| Kashmire |
I really like the source-follower idea. I agree if the voltage gain can be provided elsewhere, then a choke-loaded source follower would work well.
Chokes:
About acquiring chokes: transformer manufactures are plenty. Many of them will provide you with the E-core and bobbin, letting you wrap your own. With a oscilloscope and some patience, you can assemble the iron bar across the E-core to achieve the right magnetic density. I’ve used thin pieces of cardboard or paper to fine-tune the airgap. The exact same core and windings can either be high inductance and low DC bias or low inductance and high bias, depending on how much distance is between the iron bar and the E-core.
Summary: many transformer manufactures can give you the materials to make an easy custom choke.
I like the schematic. 50mH may be enough for full-range sound. If you operate a tri- or bi-amped system, the midrange and tweeter amps could use 50mH or less. The woofer or subwoofer amp could be push-pull. These amps can be custom-designed for systems. I run a tri-amp dipole system with good effieincy. A good 10-watt amp is enough for the midrange and tweeter, which is why I’m looking at the choke idea. |
|
|
| Fuling |
dhaen: There must be someone on this forum who eats female dogs, I just lost one too:D
Kashmire: Biamped dipoles, interesting! That´s how my system is going to be as soon as everything gets in place. At least I hope so...
I believe the schematic I posted is more similar to pic no. 2 on John Broskie´s site, the buffer version.
I´d rather use tubes and stepdown transformers than a 1:20 stepup at the input, though. |
|
|
| Fuling |
I just read something interesting at John Broskie´s site:
"This amplifier, unlike the previous MOSFET amplifiers, must be run in strict class-A, so the idle current will be quite high, at half of the peak current demand. How do we find the peak current demand? This amplifier is effectively the equivalent to a normal source follower amplifier with +/- voltage rails twice that of this amplifier’s single rail; the magic of inductors. In this case, we can assume that 12 of the power supply’s 15 volts will be deliverable into an 8W load, so 24 volts of peak voltage would require 3A of peak current (as one terminal goes up the other goes down). Thus, an idle current of 3A is needed (24V / 8W), 1.5A per MOSFET."
This seems to indicate the difference between push pull and balanced SE operation: Since I´m going to use separate , airgapped chokes for each "leg" of the amplifier I´ll need twice the bias current, 3A per mosfet instead of 1,5A if a centertapped choke is used.
Do I make any sense here? Maybe I should wind myself a pair of CT chokes... But on the other hand, I love SE amps:hot: |
|
|
| Kashmire |
Here’s a schematic … I have some 5965 and 7788 tubes here with an LL1660S/18mA interstage transformer. I’m planning on building a tube amp with these parts in March. I don’t have my output transformer yet, so these things are laying around unused …
Don’t focus on the driver stage. I’ll probably use a pentode (in triode mode) for gain around 40 and low plate impedance. These schematic I attached here shows the driver in parafeed, but I’d probably use the LL1660S interstage as single-ended running at 15mA.
Knowing that the driver will have some voltage gain, the MOSFET can run as source follower. Using a 50 mH choke with 2A bias current, this amp can run as low frequency as most good speakers – especially transmission lines and/or folded horns. I probably wouldn’t use +20V. That may be a bit high.
Don’t pick on the 18k and 2k resistors. I only used those for an example. The actual values would depend on the MOSFET and the desired bias current.
The output capacitor will probably bug some people. Nelson Pass doesn’t seem to mind too much. His Jan 2004 article explicitly says to worry about the transistor before worrying about passive components.
It’s easy enough to get rid of the output capacitor. My next post will include a schematic. |
|
|
| Kashmire |
Here’s the transformer-coupled source-follower single-ended tube hybrid MOSFET amp with a DC servo loop to eliminate offset at the output.
I didn’t draw the tube drive stage. Refer to the tube forum or previous posts on this thread to see tube drivers.
Here’s how this circuit works: the op-amp is the servo loop that keeps the inverting (-) input at ground potential (by referencing ground at the non-inverting input). It does this by modulating its output to change the MOSFET’s bias. In this case, if the op-amp maintains about -5V, the bias at the MOSFET gate is -11V, or +4V from gate to source. Actual voltages and resistor values would depend on the MOSFET’s characteristics.
The bias current flows through the DCR of the choke, which causes a voltage drop. If the DCR is 3 Ohms, then a +3V power supply is needed. If the power supply voltages change a little, then the bias current changes to maintain 0V at the output.
In summary: the op-amp controls the MOSFET's bias current so that exactly the right amount of voltage is dropped over the DCR of the choke to maintain 0 volts at the output. This means (worst case) that a DC offset could occur at the output for 1 second.
The slew-rate of the servo loop is managed by C, which should be sized below the lowest frequency of the amplifier. In this case, 1Hz would work.
The signal is applied by an interstage transformer connected directly to the gate (no gate-stopper resistor is shown). The amp is configured as source-follower by the AC coupling capacitor Cac. This capacitor couples the interstage transformer to the drain of the MOSFET, making it operate in source-follower mode.
So how does the amp deliver the positive half of a waveform? As the MOSFET cuts the current off, the choke reacts and releases its stored magnetic energy in the form of positive current, resulting in positive voltage at the output. Here’s where the frequency limitation comes in: if the choke is too small, it doesn’t have enough stored magnetic energy to create a low-frequency waveform. The lower the frequency, the more stored energy in the choke’s magnetic field you need.
This amp may not be practical to build, due to the crazy +3 volt power supply requirement, but certainly feasible in theory. If you wind your own power supply transformers (or you use switching power supplies), this amp is very feasible. |
|
|
| Fuling |
Looks good, but I can´t qute figure out how it´s going to function as a source follower. Looks more like a Zen to me.
Does it have anything to do with the fact that the input is connected between drain and gate instead of gate and ground? |
|
|
| dhaen |
The problem with "losing" any cap by providing bias, is that it' hasn't been lost at all. It's just been moved fron the signal line path to the signal return path (PSU).
No critisism intended :) Great ideas coming out here. |
|
|
| Fuling |
Pretend that we leave balanced SE and PP aside for a while and look at some basic SE designs.
Drawing A shows a "regular" choke loaded Zen design that has ~twice the efficiency of a CCS loaded design. Interesting.
The big drawback is that the PSU has to be well regulated if we want to keep the bias reasonably stable. Temperature drift might be a *female dog* too.
Drawing B should be quite insensitive to PSU variations thanks to the resistor from source to ground. The problem is that this resistor introduces us to current feedback and the high output impedance which is related.
The voltage gain here would be
Rload / Rsource, which tells us that the voltage gain depends on the speaker impedance.
Might work good with some speakers (open baffles?).
Drawing C has what I think can be called mixed feedback, we have both voltage and current feedback. With the right combination of resistor values we should be able to achieve a quite high but well defined output impedance and still keep bias and temp drift under control.
This might be an attractive solution for certain speakers, I´m thinking dipoles again. |
|
|
| Tyimo |
Fuling! Magura!
| quote: | | BTW with chokes instead of resistors in the SOZ the gain probably goes way up so some kind of negative feedback might be necessary. Not attractive. |
I don't understand why can not use chokes for SOZ without to need negativ feedback for the circuit?
In the: Pass Son of Zen w/current sources (etc.) theread GRollins and Mr. Pass wrote it is possible:
"Yes, inductors would work fine. The single-ended tube folks do it all the time. It works even better without the resistor--simply replace R1&2 with really <i>big</i> inductors. The reason I haven't pursued it is that you can't just walk out and buy large value, high current inductors, and for some reason most people seem reluctant to wind their own, judging from the threads on the topic here."
"The inductance will determine the frequency response. It's a simple filter. A smaller value inductor will give you high frequencies only--perhaps good for a tweeter amplifier in a biamp system. The larger the inductor, the lower the frquencies the amplifier will reproduce. Kind of a built-in active crossover. Unfortunately, the amplifier is still trying to reproduce the lower frequencies, so you lose the benefit of less strain on the amp. Still, it will work quite nicely if you want to follow that path.
To keep track of the idea, remember two things: 1) the larger the load on a gain device, the more gain (other factors being equal), and 2) an inductor is, in a sense, a frequency dependent resistor. At DC (in other words a really, really low frequency...F=0), an inductor is close to a dead short, limited only by the resistance of the wire itself. As the frequency climbs, the impedance increases and with it, the gain. More inductance will lower the frequency at which the gain starts to climb. No inductor at all, i.e. connecting the Drains of the MOSFETs directly to the rail, will not hurt anything (as long as you don't exceed the limits of the devices) but will produce no gain.
An air core coil is a good idea because air doesn't saturate when the magnetic field gets strong. Iron core inductors are smaller but the price you pay is in performance.
A 1 Henry air core coil is a pretty large coil in every sense of the word. Hint: It'll be cheaper if you wind your own.
Go for it."
I would like to build one SOZ amp with chokes and I would use inductors made with torroid core ca 1 H, but I don't know how big resistance I need. May be 7-9 R? So I would get a "very inductiv wire wounded power resistor" :-)
Well, my question is how can I calculate the rail voltage and transformer VA rating for a Choke loaded SOZ???
Thanks! |
|
|
| Kashmire |
True, most amplifiers have capacitors in the output path – the PSU capacitors. For example, in a class AB amplifier, the AC component has to come from and go somewhere. That’s why big PSU caps are useful. They store energy in the capacitors can be delivered to the output.
However, with some class A amplifiers – such as the Zen – the PSU capacitors don’t play a part. The AC component of the output is a modulation of a steady-state current. Therefore, the power supply delivers steady-state and doesn’t see any AC component.
Indeed, the schematic I posted involves the PSU capacitors in the output path. The output is referenced to ground, which means the signal has to find its way to the +3V and -15V rails, which will be through the PSU capacitors.
I happen to like output coupling capacitors or transformers. I’ve blown too many direct-coupled tweeters with amps with DC offsets. That’s why the Zen amp appeals to my as a tweeter amp. Only my tube amp offers the same appeal. |
|
|
| Magura |
| quote: | Originally posted by Tyimo
Fuling! Magura!
An air core coil is a good idea because air doesn't saturate when the magnetic field gets strong. Iron core inductors are smaller but the price you pay is in performance.
A 1 Henry air core coil is a pretty large coil in every sense of the word. Hint: It'll be cheaper if you wind your own.
Go for it."
I don't know how big resistance I need. May be 7-9 R? So I would get a "very inductiv wire wounded power resistor" :-)
Well, my question is how can I calculate the rail voltage and transformer VA rating for a Choke loaded SOZ???
Thanks! |
In the beginning of the thread we all agreed to the number of 50mH as being adequate, i dont see why we should not stick to that...??
If you wind an inductor with an internal resistance of 7-9 ohm you are back where we started....or not??
But just in case you wind an inductor with 8 ohm resistance, the rail voltage and so on would be the same as for a regular SOZ.
I can get magnet wire for approx. 10 usd (65 DKK) a kilogram, give and take a bit according to dia. , so winding a few air core chokes would be no big deal and not particulary expensive (got the right equipment for winding...usually used for windings for electro motors).
If anybody is interested i can find out how much it would cost to make a few 50 mH air core chokes, able to carry 200W.
Magura |
|
|
| Kashmire |
Here’s a choke-loaded source-follower Zen with some stabilized bias.
The zener diode (buffered by the 5k resistor) establishes a stable +10V reference across the bias resistor network for a stable Vgs. Since zeners are well-known sources of noise, the zener should be buffered with some caps - maybe even a little choke! (o.k. I'm getting out of hand. Where's Circ when you need 'em?)
Since we’re trying to control bias current, not gate voltage, a better way would be creating a stable reference for current. A DC servo loop could be used to control bias current using a reference. Just as a conversation piece, I’ll cook one up.
The signal is applied to the gate (through a gate-stopper resistor) with the secondary of the interstage transformer. The signal path is completed to the drain by the coupling capacitor Cac. The capacitor also serves as the local MOSFET drain-to-gate feedback path. This is very similar to the feedback of a triode-connected pentode (vacuum tubes). The feedback consumes all of the available gain, so the transistor operates with unity gain.
This forces more demands on the driver stage, which should produce a gain of about 15.
This is similar to my discussion on the push-pull thread: http://www.diyaudio.com/forums/show...15&pagenumber=2
Additional feedback is provided by a source resistor (0.25 Ohms in this example). An added bonus is increased stability.
I’ll be running some SPICE simulations of these circuits over the next few days. If someone beats me to it, post your results! Simulations should solve the 50mH vs. 1H debate, too. |
|
|
| Kashmire |
OK, this is my last post for now. Here’s the choke-loaded source-follower Zen with DC servo bias current control.
The op-amp compares a 0.5V stable reference on the noninverting input (+) with the voltage across the source resistor Rs at the inverting input (-).
At 2A bias current, 0.5V will form across Rs. The op-amp controls the voltage on the bias resistor voltage divider, forcing the MOSFET to maintain 2A.
The 0.5V stable reference is created using a 2.5V zener and a voltage divider. Of course, the zener should be buffered by some filtering capacitors.
Since the op-amp has to work at very low voltages, a negative power supply for the op-amp may be needed. One or two volts would be enough. |
|
|
| Fuling |
I´m glad to see that this thread is still active!
Unfortunately I don´t have time to look closer at the drawings and interfere with the discussions right now, ´cause tonight I have some prototype breadboarding to do.
The topology will be something like drawing C in the last picture I posted, but the resistive voltage divider for the voltage feedback will be replaced by... something else:D
I´ll report tomorrow or as soon as I have anything to report. |
|
|
| Tyimo |
| quote: | | In the beginning of the thread we all agreed to the number of 50mH as being adequate, i dont see why we should not stick to that...?? |
O.K. My original plan was to use a torroid core inductor for the choke loaded SOZ ,becouse I read somewhere that it will not or not so much saturate when the magnetic field gets strong like the classic Iron core inductors. So I would get very low resistance ca.2 Ohm. If I would use an 1H air core, than I would get with 16 gauge wire 5.2 Ohm or with 18 gauge 7.6 Ohm, but both can handle over 300W and I would get the "super inductive power resistor" with much heat like in the original circuit.
I am not sure the 50mH would be enough for a "fullrange amp". Mr. Pass built one also with 1 H load using a full roll of 16 gauge MWS magnet wire. ( with high resistance like a power resistor, but 40 -50 % efficiency)
"tube amps usually use about 100mA, and chokes are relatively easy to make for lower currents. The Zen amps use an order of magnitude more current, which means the chokes have to be that much larger. Pass Labs probably selected not to use chokes because it may be financially and commercially infeasible."
My questions about choke loaded SOZ:
-Should I use the inductors instead of the all power resistors from 1 to 6 or only of the 1-2? (numbers are from the original article )
-Need I make the 1 H inductor resistance as low as possible or it is not important.
-Should I use torroid core inductor?
-How can I calculate the bias and the needed PSU ratings, heat sinking?
Thanks |
|
|
| Circlotron |
| quote: | Originally posted by Tyimo
I am not sure the 50mH would be enough for a "fullrange amp". |
The 80mH ones on mine are just fine. If you want to avoid an output coupling cap, if the choke dc voltage drop is less than ~300mV or so then just wire the speaker right across it.
Have a closer look at mine with the choke in the source lead. The 325 mV @ 3.5 amps across it is also used for current servo sensing. It is very very tame.
/Circlotron - reading every line in the post, but has to bite his tongue to stop butting in too much. |
|
|
| Magura |
| quote: | Originally posted by Circlotron
/Circlotron - reading every line in the post, but has to bite his tongue to stop butting in too much. |
Hmm...maybe its just me, but id prefer loads of "butting in", since this thread have changed from "travelling to a destination" to a sightseeing :)
We see a lot, but get nowhere.
Is there anything gained by just replacing the r1 and r2 of the SOZ by chokes of 80 mH?
How big would the resistance of the choke have to be?
If down to 80 mH, i see no reason to make EI chokes, an 200W air choke isnt that big.
Magura |
|
|
| Magura |
| quote: | Originally posted by Magura
Is there anything gained by just replacing the r1 and r2 of the SOZ by chokes of 80 mH?
|
Or even better a 50 mH choke
Magura |
|
|
| djk |
http://madisound.com/3way.html
14.5mH, 1R13, USD$4.25
15W/8R would be 2A bias, less than 5W in the coil, 2.25V drop, 88hz is -3dB for 8R.
They have chokes of this value with lower DCR for more money, I guess the real question is how much does the inductance change at 2A DC?
As it looks, this might be good for an amplifier used with a 200hz/24dB crossover. |
|
|
| Circlotron |
| quote: | Originally posted by Magura
Is there anything gained by just replacing the r1 and r2 of the SOZ by chokes of 80 mH? |
What power output is SoZ? 10 watts or so?
In general principle, particularly with bigger amps you get half sized power supply, half sized heatsinks, half sized electricity bill.
The power supply and heatsink thing can mean big $$$ saving. |
|
|
| Magura |
| quote: | Originally posted by Circlotron
What power output is SoZ? 10 watts or so?
In general principle, particularly with bigger amps you get half sized power supply, half sized heatsinks, half sized electricity bill.
The power supply and heatsink thing can mean big $$$ saving. |
A SOZ is scaleable from around 5W to 50W, though more than 25W isnt practical dur to powerconsumption (more than a regular mains outlet can handle (230V 13A)).
I just dont understand how i can replace the 8 ohm bias resistor with a 2.5 ohm inductor...something must go wrong....at least the gain must go through the roof...no???
Magura |
|
|
| Circlotron |
If the choke had zero resistance you should reduce the dc supply voltage by whatever was normally dropped across the 8 ohm resistors. With a 2.5 ohm choke it should be reduced by somewhat less. The bottom line is, no matter what the configuration, at no signal the source-drain voltage should always be the same.
A source follower is WAY less hassle. An interesting way that Fuling suggested was to have two source followers running antiphase and the speaker connected between the sources. It is good because
1/ there is no choke voltage drop applied to the speaker.
2/ even harmonic distortion from the square-law charactistic of the mosfet is cancelled out
3/ you can use two smaller chokes.
You can't use a single centre tapped choke for this because then the two opposite direction currents make you lose the effect of the energy stored in the airgap. Well, you *could* use a single centre tapped winding with no airgap but then it becomes a transformer and the imperfect coupling between the winding halves introduces a whole new set of problems.
Try a source follower cct and see how you go. It's reeeeaaaalllly easy.
One other thing, see how well a source follower cct goes for you; it goes great for me.
BTW, did I mention to try a source follower.
New mantra - source follower.
PS. Source follower. |
|
|
| Circlotron |
| quote: | Originally posted by Circlotron
2/ even-harmonic distortion from the square-law charactistic of the mosfet is cancelled out |
I suppose it would in the SoZ setup too.
I think the world needs a source-follower SoZ. |
|
|
| Fuling |
Source followers... Yep, I´m in!
I´ve spent some hours this weekend trying to get a Zenlike creature with mixed feedback to work good but I must admit that I failed. My secret ingredient here was a low voltage triode instead of the resistive network that connects drain - gate- ground.
What I essentially did was to insert a cathode follower in front of the Mosfet but instead of connecting the the plate to the positive rail I wired it to different taps on the choke to get voltage feedback.
BTW there was some current feedback too in the circuit.
I said previously that this idea failed, but it didn´t fail completely.
The problem was that I only got 1W out before clipping when the voltage feedback was connected, and when I removed it (plate directly to positive rail) I got almost 4W.
To make it short: I lost half the swing. Bad.
Quite sad I must say because except for the big loss of swing rthe circuit worked quite well, I did some measurements with different ratios of voltage and current feedback and the results was very interesing.
Bottom line: Source followers... Stable, linear, easy to work with and they require no feedback. Gain can be had somewhere else. |
|
|
| Tyimo |
What is with Fulink's original idea in post 7: the choke loaded IT-coupled balanced SE tube/mosfet hybrid action?
Is it work? |
|
|
| Fuling |
Haven´t tried it yet. Don´t know if anyone else has before either, but I guess someone has. Wait a minute, doesn´t Pathos use that topology in their output stages?
Anyway, I will build it someday when I have the time and money to spare, right now there are are a few other projects in the pipeline. |
|
|
| Magura |
So, would this mean that i can reduce all the 8 ohm resistors in the SOZ to the resistance of the respective choke, and lower the supply voltage to match the change of resistance??
Magura |
|
|
| Henrik |
| quote: | Originally posted by Circlotron
I think the world needs a source-follower SoZ. |
This is an very interesting thred this one.
I have simulated this XSOZ-Follower with chockes instead of currentsinks.
It looks verry promising in the simulator, and this is only the first shot.
Regards |
|
|
| Henrik |
It seems as wee can get away with only 1 mH chockes besause the xfeedback "multiplys" the inductors in the closed loop, in open loop the -3db low end roll off is at 10 Hz or so.
It seems a little difficult to controll the bias, in the scematic i posted, the current is around 200 A:hot:
R101 and R14 controlles the bias current, they shoud be between 2 and 2.2 Kohm, but then unfortunately the gain gos way down, so wee are not quite there yet. |
|
|
| Fuling |
I think 1mH is way too low to use in this kind of circuit.
At 20Hz the reactance is only 0,1256 ohms, so if we put a 20Hz 15V peak signal across that choke the peak current will reach 120A.
I think the minimum inductance can be calculated such that the inductive reactance (XL= 2*pi*f*L) at the lowest frequency to be reproduced is equal to or higher than the load impedance.
For 8 ohms load that would be ~60mH (30mH for 4 ohms load).
When an amp is running in balanced mode such as the SoZ or the circuit I posted in the beginning of this thread each half of the amp "sees" half the speaker load, similar to what happens when two ordinary amps are bridged.
This means that instead of using one >60mH choke for an 8 ohms load we can use two 30mH chokes instead.
As Circlotron already has pointed out there are some benefits related to balanced circuits: The possibility of direct coupling to the speaker and cancellation of second order harmonics.
I can think of one more: Constant current draw from the PSU.
Thinking of it, the PSRR must be great too. |
|
|
| Henrik |
| Without feedback you are quite right (like in the attached file), but that was not my point, it seems you ignore tehe feedback in your calculation which is ok for nonfeedback circiuts. |
|
|
| Kashmire |
Henrik! Using feedback to correct for an undersized inductor?
My feedback theory is make the circuit operate correctly without feedback, and then apply the minimum amount of feedback to correct for unpredicted or inherent nonlinearities.
Feedback forces the lower-order distortion by-products up. Feedback won’t eliminate the distortion – because feedback has a time constant, distortion crops up at a higher frequency. Usually this is in the upper (non-audible) band, but it creates undesirable intermodulation distortion.
To operate an amp with a 1mH choke, an enormous amount of feedback must be used to correct the frequency response. No wonder the currents were so high!
Make the amp work without feedback, and then add a bit in later to correct for the MOSFET distortion. |
|
|
| Fuling |
Exactly.
Even if the frequency response can be straightened up by using feedback, the current peaks will still be there.
What we really want between source and ground is something with zero DC resistance and infinite AC impedance, and the impedance of a 1mH choke is far from infinite. |
|
|
| Fuling |
Chokes, anyone??
Have a look at www.hammondmfg.com
Ex:
159ZC 60mH 2A 0,7 ohm
159ZE 28mH 3A 0,43 ohm
193V 150mH 3A 1,0 ohm
195T5 100mH 5A 0,64 ohm
Remember that chokes can be wired in series or // just like resistors. |
|
|
| Henrik |
Hi all
My point is to get the best sound from an amp, not only in therory but as a practical matter, where you ears makes the final judgement. My goal isn´t to correct for an undersized inductor no matter what.
Distortions figures don´t always make me nerveous, i have worked with x-feedback for two years now, an if proberly implanted in the right circiut it really shines. The scematic i suggested includes not only feedback but also x or supersymmetry, since i have had so good realworld expieriences with xing the SOZ and BSOZ. So X-ing is my goal, and all i can say is, that you might get away with a smaller inductor with regard to the bandwith measured, not nessesarily to the sound, i agree on that one. But the question is, does the inductor need to be as large as 50mH to 1H in an X-circiut? May be may be not, i haven´t tried or heard.
| quote: | Originally posted by Fuling
What we really want between source and ground is something with zero DC resistance and infinite AC impedance |
Wishfull thinking.
I haven´t said that i have all the aswers to this amp - i have just been thinking loud - and the problem about the bias isn´t solved among others, and may be it can´t. I will give it a try later
Regards |
|
|
| djk |
I tend to agree with Bob Pease (National Semiconductor) that simulation programs (Spice) are a crutch and hinder learning.
Maybe if they came with a sound-effects generator, went BANG!, and emanated smoke at the right time....
No amount of feedback will change the amount of energy stored in a reactive component.
The circuit in post #51 needs energy storage in the chokes. It also needs a level shift device (coupling cap, folded cascode, interstage transformer, whatever) between the diff inputs and the outputs. Then the outputs need bias. |
|
|
| Henrik |
| quote: | Originally posted by djk
I tend to agree with Bob Pease (National Semiconductor) that simulation programs (Spice) are a crutch and hinder learning.
Maybe if they came with a sound-effects generator, went BANG!, and emanated smoke at the right time....
No amount of feedback will change the amount of energy stored in a reactive component.
The circuit in post #51 needs energy storage in the chokes. It also needs a level shift device (coupling cap, folded cascode, interstage transformer, whatever) between the diff inputs and the outputs. Then the outputs need bias. |
No - the energy stored cant be cahnged by feedback, i guess that is why it needed 200A in the sim.
The circiut in #51 works fine with resistors instead of chockes, and i can´t see why a levelshift device should be nessesary when using chockes, i will take a look into that suggestion.
djk
Come on - i thought we were here to have fun and learn - not to insult each others intelligence. |
|
|
| Circlotron |
| quote: | Originally posted by Henrik
It looks verry promising in the simulator, and this is only the first shot. |
I've got a simulator but I never use it because I can't figure it out.
:rolleyes: Anyway, could you try these two suggestions?
1/ Replace R101 & R14 with a single-centre tapped transformer winding with the ends going to the gate resistors and the centre tap going to a somewhat reduced supply rail.
2/ Replace R8 with a proper current sink.
Doing either or both of these should make the input fets at least, very symmetrical wrt each other. |
|
|
| Fuling |
Of course, since there are no perfect components.
A real world 50mH choke that can handle 3A DC seems to have at least half an ohm of DC resistance, so the voltage drop is an issue.
Is the DC resistance a bad thing? Not necessarily, since we can use it instead of source resistors to stabilize the bias point. |
|
|
| Henrik |
| quote: | Originally posted by Circlotron
I've got a simulator but I never use it because I can't figure it out.
:rolleyes: Anyway, could you try these two suggestions?
1/ Replace R101 & R14 with a single-centre tapped transformer winding with the ends going to the gate resistors and the centre tap going to a somewhat reduced supply rail.
2/ Replace R8 with a proper current sink.
Doing either or both of these should make the input fets at least, very symmetrical wrt each other. |
I will try tomorrow. |
|
|
| Circlotron |
| In 2 hours and 2 minutes it will be tomorrow here. i.e 2004. :) |
|
|
| Henrik |
Okay!!!
It woun´t be before 24 hours from now, and only if i can avoid the useually :dead: hangover.
But else i am eager to try it out in the sim.
Happy new year to all
Regards. |
|
|
| Fuling |
11 hours and 26 minutes left of this year here.
Happy new year to everyone, and don´t choke on the champagne! (pun intended) |
|
|
| Pan2 |
Someone did mention something about the Zen amp not having the PS in the signal... it´s not so I believe. Both the Zen and Aleph amp has the PS in series with the signal. The SOZ has not.
The CCS only delivers current for one 1/2 wave, the other half is via the caps in PS.
I really like the idea (and have been thinking on it for a year) of a choked loaded balanced/bridged source follower without feedback. Using the preamp for voltagegain.
I see the greatest benefits for this circuit in the high frequencies, makeing it easier regarding the choke, which I assume can be an aircore... maybe a foil.
Never realized the problem with extremly high currents at low frequencies though, where the reactance of the choke is low. Seems like a highpass earlier in the path would be necessary???
Happy new year!!!
/Peter |
|
|
| Nelson Pass |
The Zen has a constant current source draw from the positive
supply. The AC output signal flies around on the ground
portion of the circuit. |
|
|
| djk |
"djk
Come on - i thought we were here to have fun and learn - not to insult each others intelligence."
I'll admit I was laughing at your sim, but you must admit it WAS laughable (200A!).
"I've got a simulator but I never use it because I can't figure it out. "
I begrudge the time lost in learning two different systems (even though I was given the software and paid to attend the classes).
I have been in the situation of the sim saying my circuit won't work, while the breadboard is sitting on the bench running.
Replace the lower CFPs with the chokes and this is what you're looking at:
http://homelf.kimo.com.tw/skychutw/...s/Hadley622.pdf
the bootstrap caps providing a CCS load for the diff pair to work into, bias adjust and DC off-set controls shown too. |
|
|
| Henrik |
Circlotron
Thanks for the suggestions, I have simulated them - almost, i used two inductors instead of a single-centre tapped transformer.
Do you have a special reason to suggest a centertapped transformer?
Speakerload 8 Ohm.
Bias current for the outputfets is 4.3A, which is better than the previous 200A:D .
The output inductors is 33mH as suggested elswhere in this thread.
The inductors at the inputstage needs to be as high as 680mH but needs only to take 50mA.
Output bias is adjusted by VR1 and VR2.
This Circiut is not ment to be ready to bulid, but more to evaluate this topology, all though i am shure it will work, but how does it sound?
I am not used to work with inductors, so i have to study that issue a some more before i take this further.
djk
Unfortunately i can´t get the pdf file you pointed out.
Regards |
|
|
| djk |
"djk
Unfortunately i can´t get the pdf file you pointed out."
Works for me, but try this instead:
http://home.kimo.com.tw/skychutw/
Then go to:
"Circuits"
And then to:
"Hadley 622"
It is a pdf file so you need Adobe (free).
http://homelf.kimo.com.tw/skychutw/...s/Hadley622.pdf
You must have Active X enabled in your security settings, I can't run TIFF files at work for this reason.
Your new circuit looks better.
The resistor plus bootstrap cap in the Hadley serves the same function as the 680mH choke does in your new circuit. |
|
|
| Henrik |
djk
Thanks for you kind answer!
I tried enabeling the activex to the lowest level. I can get the page http://home.kimo.com.tw/skychutw/ but still can´t get the pdf.
I have the reader installed.
I will give it an other try later.
I have found some litreture about bootstrapping on the net, which i will read when i get the time to do so.
To day i will finish my new xsoz.
Thanks again
regards |
|
|
| Henrik |
djk
I just got the Hadley622.pdf.
Thanks |
|
|
| Circlotron |
| quote: | Originally posted by Henrik
i used two inductors instead of a single-centre tapped transformer.
Do you have a special reason to suggest a centertapped transformer? |
Yes. It guarantees that as Q101 drain goes up for example, Q6 drain goes down by exactly the same amount. With two separate inductors they not locked to each other but are free to move independently which is not good. :whazzat: |
|
|
| Henrik |
Circlotron
I got it, makes sence.
Thanks |
|
|
| Henrik |
| I have remooved VR2, C3 and R8 from the scematic in post #71, we need only one biasadjustment. |
|
|
| Henrik |
Thanks for the link, very educational.
Regards. |
|
|
| stefanobilliani |
In that link , there are at least 2 pics were the gate of the gain mosfet is marked at 1.5volts to ground. How is it possible ?
Does not a mosfet require at least 3.5 volt (gate - source) in order to turn on? |
|
|
| stefanobilliani |
:angel:
Thank you very much Henrik.
Happy new year to you too. |
|
|
| Kashmire |
Here’s a choke-loaded follower Zen with a tube driver – no interstage transformer this time.
This circuit lets each part perform their best function: the tube is a voltage device, therefore, the tube performs voltage gain. The MOSFET is a current device, therefore, the MOSFET performs current gain.
I’ve shown some grid-stopper (or gate-stopper) resistors. There may be a few parts missing, such as a gate protection zener diode.
Now here’s the time for me to look like a real noob:
- Since the MOSFET is working in follower configuration, all of the voltage gain must be performed by the tube. How much voltage gain is needed to drive the gate? An IFRP240 has a transconductance of 6.9, so only about 2 or 3V peak-to-peak are needed at the gate? The choice of the triode will be dependant on the amount of voltage gain required.
- What’s the input impedance at the gate? A better question (showing how long it has been since I’ve worked with FETs), how is this calculated?
- The gate will probably have about 1000pF capacitance, so the driver stage should be about 1k output impedance to maintain 100kHz. Better look at those triode specs...
I’ve got everything to build this amp except for the choke. I plan on winding one this week. |
|
|
| stefanobilliani |
Kashmire,
I also need the inductor . Can you show up how to wind it in pratical?
I have got a pair of E83CC , can they do the a decent job in this application , or are they best employed in pre-pre cicuits such phono stage? |
|
|
| Circlotron |
| quote: | Originally posted by Kashmire
- The gate will probably have about 1000pF capacitance, so the driver stage should be about 1k output impedance to maintain 100kHz. |
That 1000pF is not the problem you might think. Just suppose the gate goes up by +10v and the source follows it by going up +9.8V. The gate-source voltage has only increased by 0.2v and it is *this* tiny voltage that pumps the gate-source capacitance, not the full 10v. |
|
|
| Circlotron |
| A 20v supply will pull +2.5 amps peak into 8 ohms. Therefore, bias the fet at 2.5 amps so the choke will be able to swing the speaker down to -20v as well. That shoud be ok for 25 watts into 8 ohms, in theory. The follower has a gain of 1 of course, so it needs 50v p/p drive on the gate. |
|
|
| joensd |
I´d love to experiment with these circuits as well.
The tube cad article and circlotrons project look very tempting.
You can easily buy inductors up to 50mH with appropriate resistance values but they also easily cost >80€.
Anyone care to explain how much inductance is needed for certain output power. For example 10W?
TIA
Cheers
Jens |
|
|
| Kashmire |
There are two values of interest for the choke:
(1) Inductance. This value will limit low frequency response. Large values mean lower frequency. I’ll be posting my choke design later this week. It will most likely be 33 to 50mH.
(2) DC Saturation Current. Too much static DC current will cause the magnetic field to saturate. The choke’s core must be able to handle the static current without saturation for the choke to retain it’s inductance. My design will be built to handle 2.5 to 3A of static DC current. |
|
|
| Kashmire |
I looked into buying chokes.
Toroidal 120mH 2.5 Ampere power chokes will cost $45/each for 10 chokes.
Dimensions: Diameter 3.9"
Height 2.0"
Weight 2.8 lbs
Not bad ... If I order 10, does anyone want some? |
|
|
| Circlotron |
Here's something I put together and debugged on Friday, that would be ok for driving a pair of source followers. You can feed just one side only and use it as a phase splitter as well, or drive it in balanced mode as per the drawing. In balanced mode it clips absolutely symmetrically. This thing has an enormous amount of gain that is ploughed right back into making the thing linear. One thing to note - the output impedance is very high and so the 1k resistors across the top choke half-primaries set the stage gain. Leave them out and the gain heads toward infinity!
The DC voltages are measured at no signal. The +62v is sort of arbitrary - it was as high as my bench supply would go. One thing is certain though, you can just wire it up and apply the voltages and away it will go. The total dissipation is about 12 watts (!) so the fets and the two voltage regs under them will need a small heatsink.
The AC voltages are measured with 2.905VAC RMS input signal just to give a general picture of the internal gain of the amp. Above this level, the output is still *extremely* linear but the mosfets start to get a little bit lazy so their sources get driven a lot harder to maintain the nice output signal. The signal is very civilised as it pulls out of clipping too.
As can be seen, the maximum output voltage is 74.66 V drain to drain. Enough to drive most things. Well, there it is. |
|
|
| Fuling |
Circlotron: I envy your ability to produce a never ending stream of interesting circuits!
Kashmire: Torodial chokes, you say? Can the handle the DC bias?
I doubt so, since they usually don´t have airgaps |
|
|
| joensd |
Kashmire, I certainly would be interested in a group buy but the shipping costs would probably be too high to be worth.
2.5A chokes? Wouldn´t that be too low in terms of saturating?
Anybody experienced how low you can go with the choke in terms of inductance.
I don´t trust my simulator.
With 33mH and a single ended follower an output of +20W still looks too good to be true. |
|
|
| Circlotron |
| 80mH is ok for 50W, i.e 28V peak, down to any real-world frequency. |
|
|
| djk |
Just like a filter capacitor not needing to be a larger value for a higher power amplifier, the choke stays the same value.
The only reason to change the value of the reactive components is if you want a different 3dB down point or change the load impedance.
In the case of the capacitor the power stored is a function of the square of the voltage stored, in the case of the inductor it is the square of the current.
80mH is good for a 3dB down point of 16hz with an 8 ohm load, doesn't matter whether it is 5W or 500W.
Unless specified otherwise assume that the inductance drops 20% at the rated DC current, or the 3dB down point raises to 20hz in the above example.
Always ask the temperature rise at rated current, it may be too hot for you. The manufacture may assume a Delta of 60*C is OK unless you tell him otherwise. This would be above the boiling point of water in a 55*C ambient (glass fronted rack shelf). In this application a Delta of 30*C might be more appropriate.
The Delta is proportional to the square of the current: so if the Delta is 60*C at 4A, then it will be 2.8A at 30*C.
The manufaturer assumes:
this is a power supply
you probably have a fan
you know what you are doing
(its not his problem if you don't)
So be sure and ask if you don't know. |
|
|
| Kashmire |
Yes, these chokes are rated 2.5A continuous current. They are specifically built for filtering DC power supplies.
I can also order 60mH / 5A chokes for the same price. To keep the same magnetic field on a similar core, I suppose a 96mH / 3.5A (ratiometric calculation) is also possible for the same price. These are from a magnetic core manufacturer.
As djk pointed out, 80mH is good enough for low-frequency response. A larger choke will give a little headroom because DIY projects aren’t always exact, and the load isn’t always 8 Ohms.
I’ve got my amplifier simulated in B2Spice. I’ll be posting a simulation and schematic soon. I expect to build the amp when I get some coils. Hopefully, I’ll order some this week.
If anyone is interested in buying some of my extras, the shipping costs won’t be too bad. Each choke weighs only 2.8 lbs. |
|
|
| Kashmire |
Here’s a choke-loaded source follower n-channel MOSFET amp.
This particular design focuses on using the best attributes of each active component. The MOSFET’s best attribute is controlling current, and the tube’s best attribute is controlling voltage. In addition, local feedback is maximized.
Therefore, the MOSFET is configured in source follower, which does not have any voltage gain. In this mode, the gate drive is severe (the gate voltage must be equal to output voltage), but the effect of gate capacitance is greatly reduced. All of the MOSFET’s transconductance is used in local feedback, which results in the unity gain.
The MOSFET is loaded with a choke, resulting in relatively high operating efficiency and allowing for power supplies down to 20V. The gate is tuned for 2A bias current. A zener diode stabilizes the gate bias, keeping the bias current immune to power supply variations. No filtering capacitors are shown, but the zener should be heavily filtered. Other required components not shown include a gate protection diode to protect against spikes over +20V at the gate. (Note: this circuit may also need a DC servo loop to further stabilize the gate bias)
The tube is employed for voltage gain. This simulation uses the 6sn7 using a 40k plate resistor at 300 volts (my prototype will use the 6c45pi). The triode tube uses the purest form of local feedback, the plate feedback, which is part of the natural operation of the tube.
This stage has a gain of 10, when combined with the unity-gain MOSFET output stage, 1 volt of signal will result in 10 volts of output. |
|
|
| Kashmire |
I’m a bit unhappy with the n-channel amplifier shown in the previous post. The load is attached across the active device (the MOSFET), which means the load is referenced to the positive power supply rail. I’d prefer the load to be referenced to ground.
Therefore, a p-channel design. Note the gate stabilizing zener is now referenced to the positive power supply rail instead of ground. Why is referencing to ground so magical? Read on …
Time to break the mental barrier: Nelson demonstrated some interesting attributes of a certain p-channel MOSFET in his latest audioXpress article. Is there a reason why a p-channel amp can’t sound as good as a n-channel amp, or is everyone just locked into thinking n-channels are the only way to go? |
|
|
| Kashmire |
This is why ground-referencing the load was so interesting: A global feedback loop has been added. This particular schematic uses a 1k input resistor and a 20k feedback resistor. The amplifier loses some gain (1 volt input results in 8 volts output instead of 10 volts output) but distortion is reduced.
Some things are missing in the simulation that I would use in “real life”: a small capacitor (pF) across the feedback resistor to trim high frequency response, heavy filtering of the gate bias zener diode, and an active current source for the tube instead of the 40k plate resistor. I haven’t shown the power supply, either. Also not shown is a DC servo loop to further stabilize the MOSFET gain bias.
I'll update the simulation with the 6c45pi model later. The 6sn7 should be beefy enough to drive just about any MOSFET in source follower configuration, but the 6c45pi is a more powerful driver with higher gain. With the 6c45pi, this amp will be very easy to drive, and the extra gain (30) can be used to experiment with feedback.
So do you actually know what feedback sound like? Build this amp and find out. The feedback can be adjusted from gross (5k feedback resistor) to negligible (1M feedback resistor). At the end of the day, I can tell you what feedback “sounds like.” I’ll report when I get my chokes … |
|
|
|
