| Experimental Design - Comments, Flames Welcome - Click HERE for Original Thread |
| mrothacher |
First, to the uninitiated: please don't build this circuit. It is only a conceptual schematic. I'd really like to hear the thoughts of your group and, if he would do me the honor, the "Big Guy" (NP)! Particularly, I'd like to figure out if there's a simple way to employ mosfets in the OP.
Thanks
Mike |
|
|
| mrothacher |
| Here's the mosfet version almost ready for alligator clips and duct tape! |
|
|
| Nelson Pass |
| OK. but I have to think about it for a while. |
|
|
| mrothacher |
| Excellent! Thanks Mr. Pass! |
|
|
| GRollins |
I glanced at if briefly. Need to get back to painting walls, etc. My first impression is: Circlotron.
If so, bear in mind that that output stage is patented by Ralph Karsten.
I'll try to look at it in more detail later.
Grey |
|
|
| reinhard |
Yes Circlotron,
I don´t kwow the content of the Ralph Karsten patent, but there are many Circlotron amplifiers out there: Einstein OTL, Audio Valve and every power amplifier from Balanced Audio Technology,
tube and transistor.
Reinhard from Germany |
|
|
| mrothacher |
Grey, Reinhard:
Thanks for your replies. Yes, I would say the OP qualifies as a circlotron. It does remind me of an old OTL tube schematic I saw in TAA. I had a look at the Karsten patent and I don't believe this circuit is is directly comparable. To my unqualified eye it is just a balanced bridge with batteries in two of the arms. In fact, the inspiration for the OP came from a 1960's U.S. Navy transistor book. I was originally trying to work-up something along the lines of a two stage "Son of Zen" and this where I ended up. I'm working on a currnet flow diagram to help clarify the circuit action in my mind. It is problematic at the moment, and I haven't figured out a good way to apply overall feedback. I wonder if th "X" feedback topology would work? Also I think I see an easy way to dynamically bias the OP, but it'll have to wait for the weekend. I really appreciate your comments. Please keep them coming.
Mike |
|
|
| mrothacher |
If I may go off topic for just a moment, I've been wanting to thank Nelson Pass for a decade or so. I built the A40 when I was in the Army, stationed in Germany, and I've been having fun with this hobby ever since, due mainly to the generous contributions of Mr. Pass. It is a terrific hobby.
Have a great vacation.
Mike |
|
|
| GRollins |
I still haven't had time to go back over the circuit but, to the extent that it is (maybe) a Circlotron, yes, X will work. Keep an eye on the phase and all will be well.
Grey |
|
|
| Joe Berry |
For anyone interested, there is a good source of early information on this topology at http://circlotron.tripod.com. The site focuses mainly on the erstwhile Electrovoice Circlotron product line, but also describes other implementations of the floating-bridge output stage concept that emerged under other names (e.g. cross-shunt push-pull, parallel-oppposed) in products and patents from the US, Finland, Japan, and Russia.
For more on the later implementations, you can do a search at http://www.uspto.gov on the Bongiorno (Sumo BJT) patent (US 4,229,706) and the Karsten (Atmasphere triode OTL) patent (US 4,719,431).
And just in case anyone reads the thread djk links to above, don't believe my statement that the Lamm amplifiers are of this type -- they're not. |
|
|
| djk |
| Its a question of the feedback loop and the bias scheme. If we refer to Pass 5,376,899 figure 7 we see the Bongiorno 4,229,706 (Sumo Model 9), figure 8 is the Sandman design from Wireless World 1971, and figure 9 is the Pass X design. It would be interesting to compare the three different feedback schemes on the same amp. My main concern is the bias. The Bongiorno bias scheme would work with the Hitachi type lateral FETs, but not the IR type. I guess a bias servo will be needed. |
|
|
| reinhard |
Hello Joe Berry,
of what type are the Lamms. When NP spokes about nine gain stages in the big L, does he mean Lamm?
Reinhard |
|
|
| Joe Berry |
djk, I think the bias scheme in the Bongiorno patent is designed to respond to changes in base current, so it is really only meant for BJT output devices. Maybe you could work up a variation on the idea suitable for MOSFETs.
The experimental circuit described in this thread is so heavily biased into class A that it may not need any thermal compensation beyond source degeneration, even with IR types. |
|
|
| Joe Berry |
| reinhard, the review of the Lamm ML2.1 by Bascom King (available from http://www.lammindustries.com) has a circuit description. I think it has five stages overall, including a vacuum tube Vas stage. The output stage is compmentary MOSFET push-pull. I don't know which "L" NP was referring to, but my guess would be Levinson. |
|
|
| P.Lacombe |
Mike,
I think that it's a problem with the non-linearity of the driver stage M1 and M2. I suggest current-feedback whith some resistors in the source of the devices, and/or global feedback from outputs to inputs.
Regards, P.Lacombe. |
|
|
| mrothacher |
Thanks for your replies! This is excellent information. I've worked up a version with overall feedback. However, the source resitors on the input differential, as suggested by P. Lancombe, might be just as effective. The bridge bias is determined by the Vgs and transconductance of the output mosfets, so heat may be a problem. I hope Joe is right about the high bias, so complex thermal compensation isn't necessary. An adjustable current source for the input pair could trim the bridge bias as well. DJK, I really liked the idea about comparing the different feedback topologies. It would be easy enough in this circuit to try no overall feedback, normal global feedback, instrumentation amplifier feedback, even super-symmetry. I've had a look at all of the patents mentioned. It is amazing how many variations of bridge amplifiers are out there. The prototype is almost ready so, we'll see if this one actually works. I can smell the ozone already.
Thanks
Mike :D |
|
|
| GRollins |
...just as long as it's ozone and not smoke...
Good luck.
Grey |
|
|
| Circlotron |
| quote: | Originally posted by mrothacher
It would be easy enough in this circuit to try ... instrumentation amplifier feedback, even super-symmetry.
The prototype is almost ready so, we'll see if this one actually works. I can smell the ozone already.
Thanks
Mike :D [/B] |
Any results yet? I know what an instrumentation amplifier is but what is i/a feedback? and super symmetry?
GP. |
|
|
| Circlotron |
Actually, is this a super-symmetry cct as done with opamps?
GP. |
|
|
| mrothacher |
Circlotron:
Super-Symmetry is the name Nelson Pass has ascribed to his U.S. Patent # 5376899. If you have a look at the patent, in figure 9 you'll see an op-amp style rendering of the invention. There's also an article on Super-Symmetry posted on the Pass Labs website. If you have an interest in balanced topologies, there was an excellent article in TAA 1/91 by Erno Borbely titled "Balanced Audio Amplifiers". I haven't yet tested the experimental design. I've made a number of changes to the schematic and I will post my changes and findings as soon as I can. I just completed the floating power supplies, so I'm hoping to test this weekend. I have tested the circlotron output stage on its own, however, and it appears to perform well, in fact it might make an interesting unity-gain power buffer albeit AC-coupled at the inputs.
Thanks
Mike |
|
|
| Circlotron |
Looking forward to seeing your schematic. This is the extent of my thinking at this point. I'm sure there are a few things with this cct that others would do differently ;) . The first set of opamps have their supply rails referenced to the same rails as the second so they too are bootstrapped. I have put rc's allover the place to damp things down a bit; I'm quite sure it would oscillate all over the place if I didn't. The idea is to fire it up and then reduce these rc's to a useable point. The bias is set by the 100mV forcing 100mA therefore 100mV across the source resistors. When the fet is on it's non-working cycle it will not turn fully off but instead have the constant 100mA through it. I think I will have to fiddle some of the biasing rc time constants so it won't try and follow the audio too much on the working half cycle of each fet.
GP. |
|
|
| janneman |
To MROTHACHER:
On the patent side, the earliest reference to this type of circuit I found is from Philips. They had an early 60's implementation with tubes, and called it PPP for parallel push-pull, as the output tubes are paralleled for AC (and in series for DC). I built it at the time with 2 x 807's.
I noticed you used separate supplies for the output stage and the pre-stage. I think you can use the output stage (floating) supply to power the input stage as well. Note that the output supplies are offset from ground by half the output voltage. So, powering the input stage from this will give it a supply that tracks the output voltage to some extend if you chose the polarity correctly. That may even be beneficial for the linearity of the thing, since it decreases common mode swing from the input stage's point of view. Are you simulating this circuit? I would be interested how that would work out.
janneman |
|
|
| mrothacher |
Circlotron:
I should have mentioned that according to the X series manuals, Super-Symmetry cannot be done with op amps. Figure 9 simply illustrates the feedback and cross-coupling of the gain blocks.
janneman:
Using the floating supplies to power the input differential would be very interesting, but exactly how to carry it out eludes me at the moment.
Thanks
Mike |
|
|
| Nelson Pass |
You could explore creation of an X circuit with
a pair of inverting "current feedback" type chips,
if such a thing exists. These would have say a
gate for the negative input and a source for the
positive input. More to the point, you would have
a power inverter on the output stage of each. |
|
|
| Brett |
Hi Nelson,
Hope you had a great holiday.
The X-Zen idea has me intrigued. Is this something you were going to do some articles on later?
Regards |
|
|
| mrothacher |
The prototype is complete. I've made a number of changes to the schematic, which I'll post later tonight. The bias is (reasonably?) stable once the heatsink is warm. A condition which may take 20-30 minutes to reach. I'll also post my initial findings and several questions (of course questions!) as well.
As currently implemented, the circuit has increased gain at LF which I can't explain and I'm quickly exhausting my brief electronics education!
Thanks
Mike |
|
|
| Nelson Pass |
The Xzen will be part of the Zen Variations series,
and I have made several different versions.
It is also my intention to release some commercial
product by the end of the year or so.
In the meantime I have no objection to you guys
working it up on your own. :) |
|
|
| Petter |
So, when building an XZen -- basically the X1000 input stage with low voltages and high current -- it would probably be opportune to use op-amps to control the current sources to reduce power consumption (each volt dropped across sense counts at multiple amps quiescent current). Such a scheme would also enhance stability, but if I know NP right, he would do it with a small Zetex BJT ...
I suspect these sources can be lifted from the Aleph low side.
Petter |
|
|
| Nelson Pass |
| op amps. ugh. |
|
|
| mrothacher |
I've tried perhaps 10 or 20 different versions of this circuit now with different resistor values and in all cases it suffers from excessive LF gain.
In some cases it has HF gain problems as well, but lowering R4 and R5 to 10 Ohms or so seems to flatten out the HF response.
The only way I've been able to reduce the LF gain is by reducing the gain to near unity, which takes the fun out of "amplification."
A couple of details about the prototype:
1. All fets are mounted to the same heatsink.
2. The measurements posted are unbalanced (- input connected to ground
3. It is built on a breadboard.
A couple of additional questions for the group:
1. How do you derive balanced outputs from your signal generators? Would a low-distortion op amp unity gain converter be the way to go? My generator doesn't have balanced outputs.
2. Anyone know how to make a balanced distortion measurement with an HP8903B?
Thanks again for your help!
Mike |
|
|
| Joe Berry |
| What are these guys doing in there? If they are for DC feedback for the output stage, it might work better to use low-value (.2-.5 ohms or so) source resistors instead. |
|
|
| mrothacher |
Hi Joe:
My thought was that R9 and R10 would allow the input differential to swing a higher voltage without disturbing the voltage referenced by R11 and R12. But my scheme must be reaking havoc on the frequency response. I've added .2 Ohm resistors to the output sources, which helps the bias, but if I eliminate R9 and R10 The output clips at 10 Vp instead of 20Vp. I wonder if I should flip the circlotron around and drive the output across the drains of Q1 and Q2. Hmmm. I hope to post some more info this weekend.
Thanks again for your help.
Mike |
|
|
| Joe Berry |
Mike, if you don't mind another suggestion, you could also try leaving R9 and R10 where they are and referencing the output stage sources to a common bias voltage source. The attached zip file shows a (very crude -- sorry!) picture of this. For the actual bias source, you could just use an adjustable voltage divider across the -20V supply, i.e. a pot and a resistor. If you get a lot of bias drift with temperature, you could then go with something more elaborate.
ADDENDUM -- You could also derive a bias voltage from the drains of the input differential pair, and that might work better once you have solved the frequency response anomaly and know how to avoid it. |
|
|
| mrothacher |
Joe: Thanks for the suggestion. It helped to determine that the frequency response problem is not related to the output section. I built a copy of the front-end only and I'm getting the same results. I'm beginning to suspect something more sinister. I think the change you suggested may ultimately be the way to go, but for now I've got to troubleshoot this front-end. I'm perplexed, there's nothing to it... I'll let you know what I find out. If I can figure it out I'm certain this thing will work quite well.
Thanks
Mike |
|
|
| Joe Berry |
| Just curious, Mike: if you are actually driving the inputs balanced, are you using a floated signal generator to do it? I could imagine this being a potential source of problems if the signal source isn't designed for balanced or floated operation. In any case, you could check for problems on the source end by measuring the signal appearing at the inputs of the diff pair. |
|
|
| mrothacher |
Joe: I'm driving the circuit unbalanced - negative input connected to ground. My audio generator is single ended - negative input connected to earth through the chassis. I was wondering if the oscilloscope might be a problem, it is a battery operated PC scope, so I assumed (perhaps falsely) that the input was floating. As a crude double check, I used my DVM to test and swept through a few frequecies. The result was the same - a long downward taper from 100Hz to 20KHz. This is my first attempt at a balanced circuit, so I'm not familiar with the proper measurement techniques. I've found very little info on the Internet or elsewhere with respect to floating scope measurements. And still, I'm not even completely sure the measurement is the problem...
Thanks for your help. When it's perfected, I'll owe you a couple of PCB's!
NP: No word as of yet. You're just going to let me wrestle with this one for awhile aren't you...! No pain, no gain. (no pun intended) |
|
|
| Joe Berry |
| Mike, is the taper on the inputs as well as the outputs? |
|
|
| mrothacher |
| No, just the output. The input looks fine. |
|
|
| Joe Berry |
| Sorry for the "20 questions," but -- have you double-checked that R3 and R8 are 220 ohms and not 22k ohms? |
|
|
| mrothacher |
Joe:
Here's a diagram of my exact test setup:
To be certain I wasn't being fooled by the scope, I connected a 22k resistor to ground (as pictured) and obsevred a 1 kHz square wave ground referenced. The result is identical to the floating output (between the drains of the fets) I must be missing something terribly simple. Also, I tried the circuit with and without the source resistors (just to be sure) to no avail. I'll check everything thoroughly (values, connections, etc.) and have another go.
Thanks again.
Mike |
|
|
| Joe Berry |
| Mike -- From your test setup and reported 100 Hz rolloff, my best guess is that R21 and R22 are actually 220K ohms instead of 220 ohms. In addition, given the "open loop" topology you're using, R1 and R2 at 10K ohms also may be rolling off the HF too soon in conjunction with the diff pair input capacitance. I'd consider replacing them with 1K values. |
|
|
| Circlotron |
I'm referring to circuit23.zip so we don't get confused. I get the gut feeling to disconnect the lower ends of R4 and R5 and run them to earth. Make R9 and R10 zero ohms.
One way to set the bias is to offset M1 and M2 gates a little negative but it will be very touchy. Probably unusable.
GP. |
|
|
| mrothacher |
Circlotron, Joe: I was thinking of using a modified current mirror in the LTP to control the output current along the lines of what Joe had suggested. I'll draw it up and show you what I mean. Not sure if it would work, but maybe worth exploring. Thanks for your recommendations. I'll try them soon.
Thanks
Mike
Joe: your guess about the 220k resistors is pretty insightful and would emulate the effect, but I've checked them. In your experience, are there any known problems with these experimentor plug-boards? I think Grey said that he built his Aleph-X on one and is not having problems. |
|
|
| Joe Berry |
Sorry Mike, I don't have much experience with plug-in boards. I use one occasionally, but generally prefer to go from simulation to PCB layout, as I think I expose more problems that way.
That said, as a service tech years ago I learned that common airborne pollutants can deposit a semi-insulating film on contact surfaces over time. So if your plug-in board or components have any age on them, I'd probably at least try cleaning and reseating the component leads. As you say, the problem does resemble a high impedance (film?) in series with the gates of the input pair. |
|
|
| Circlomanen |
I have designed and built a two stage solidstate circlotron amp without any couplingcaps. It uses 2SK170GR in the input/feedbackstage and IRF 9530 in the outputstage.
Its only giving me about 14 watts a channel but with more heatsink and bigger transformers i could get 50-60 watts.
It souds way better than my zen version 4 i tried prior too this design.
Right now it has current feedback, but it is easely changed too supersymmetry feedback. |
|
|
| mrothacher |
Hello:
It has been a long time since this thread has seen activity. I haven't had time to experiment with this one. I'd love to see your schematic, if you're able to post it.
Mike |
|
|
| Tom2 |
Your link sends me to main page.
I looked through the web site and found it?
Is this it?
Hopefully the attached schematic works. |
|
|
| Tom2 |
Small correction:
I said "hopefully the attached schematic works", what I meant to say is "hopefully attaching the schematic works".
I am thinking of designing and building a circlotron/susy soon.
I am thinking of using a balanced line zen type or reflected cascode front end, then drive a balanced circlotron output. Probably use local or global feedback. I'm shooting for 100 watts class A into 8 ohms. I'll try low power version first just to see if my ideas work.
I have a singled ended circlotron design idea using an aleph type current source. It is based on the Penultimate Zen circuit. I don't know if I'll build it.
I also have a design idea based upon the Bongiorno 4229706 patent. It uses CFB amps for the SuSy topology.
I'll post the schematics if your interested.
Tom |
|
|
| Circlomanen |
Thanx for the help whith the picture!
I would very much like to se yuor schematics. Im very intressted in designing and building circlotrons.
My current design is that on the picture, and it works very good.
Its a klass a/b design and i use 4 IRF9530 /chanel.
ts by far the best sounding amp i ever had. I will some day modify the feedback to susy-feedback and crank up the quisent current in both stages.
(Sorry my English sucks, but i hope you understand what i mean.) |
|
|
| Tom2 |
I am trying to figure out your circlotron circuit.
This schematic is just and idea.
It is based on the Borgonio patent.
Please excuse me if my references to intellectual property are incomplete. |
|
|
| Tom2 |
Please excuse the hype on the schematic.
Note: I have not built or tested any aspect of this schematic.
The aleph section could be replaced with a ordinary current source or even a resistor.
A low power version or preamp could be built with IRF610's for example, with the aleph current source replaced with an ordinary current source.
A problem with this circuit would be the output offset voltage.
Of course these ideas come from tube circuits.
Tom |
|
|
| Circlotron |
| quote: | Originally posted by Tom2
This schematic is just and idea.
It is based on the Borgonio patent. |
Tom2, I think there is a drawing error in your :cool: circuit.
I think the outer ends of R8 and R9 should both go to the earthed junction of V3+V4.
Hmmm... the resistor Rx that connects U1- to U2- can't be left just floating and connecting two inputs; it will have to have some dc connection (even if via a high value resistor) to somewhere.
Also, what is the reason for R6/D1 and it's neighbour? I can sort of see what it is going to do but I can't see the reason yet. |
|
|
| Tom2 |
Circlotron thanks for asking.
The circuit is a very idealized representation.
Things like compensation and stabilty are not even addressed.
From what I understand of the circuit, at the quiescent point(no signal input, inputs at zero volts), the outputs across the load sit at zero volts with respect to the ground point of the voltage supplies V3 and V4 for the CFB amps(via feed back action of R2 and R3 or R4 and R5). Thus a potential exists across the resistors R8 or R9 equal to the voltage of V4, thus current flow from the output nodes to the negative supply terminal(V4) through each resistor is equal to V4/R8 or V4/R9. Since the circloton voltage sources (V1 and V2) and output transistors(Q1, Q2) form a closed loop, this current must come from the output of the U1 and U2, thus this is the base current for Q1 and Q2, thus sets the circulating bias current for the circlotron loop. The circulating current is thus equal to beta of Q1 or Q2 times current through R8 or R9.
The negative inputs are low impedance inputs of U1 and U2 because they are CFB amps. You might be right there should be resistors to ground. I was looking at the schematic in
http://www.diyaudio.com/forums/show...9639#post129639
Post#8 in the Monolithic SuperSymmetry with Current Feedback thread for example.
Also resistors could be added from the positive inputs of U1 and U2 to ground if the inputs are left to float(as seen in other susy circuits).
I think???, the resistor, diode, R6,D1, increases the circlotron circulating current when the corresponding node goes positive with respect to ground. This prevents Q2 from cutting off(since Q1 is conducting more and Q2 is conducting less to force current through the load because of the amplifying action). R7, D2 does the same for Q1 for the "opposite cycle". I thought of leaving them out initially for simplicity. I'm still unclear about their operation
If mosfets are used for output devices then the biasing scheme is obviously different.
I like playing around with circuits in an abstract way, with sometimes no attention paid to if the circuit could actually be built, work or sound good. Sometimes sort of a copy, cut and paste
method to see if different ideas can be combined; not even completely understanding a circuit.
I have many ideas for susy circuits and susy-hybrid circuits.
Do people on this forum find "fuzzy" audio circuits like this one useful? -- or do circuits need to be more diy-able?
Tom |
|
|
| Circlotron |
| Fuzzy circuits are great. Probably also a note to say they are such, so if someone doesn't realise but still wants to go ahead and learn stuff -> :bawling: then they are forewarned. |
|
|
| Tom2 |
Here is a fuzzy SuSy circuit you might find interesting.
Again it is an idealized circuit.
I would few it in an universal amplifier context.
It is not a circlotron.
Gotta take a break from those circlotrons.
To much mind candy.
Tom |
|
|
| Tom2 |
Here are some more SuSy circuits. They use opamps. (I know---girlie chips)
I would call these circuits "inies" as compared to some of the Nelson Pass gain clone supersymmetry design "outies" (e.g. GC-SS-4a).
IMHO the XGC2 circuit also shows a possible way to modify (for example the GC-SS-2a), so the common mode signal input impedance does not look like a negative resistance.
Also something I found recently.
Look at US patents 6,741,129 and 6,717,467. High frequency stuff though.
Tom |
|
|
| Tom2 |
Remove R5, R6, R7, R8 for a SuSy circuit.
Remove R1, R2, R3, R4, R9 and connect the bases of Q1 and Q4 to ground for a CMAC bridge.
Tom |
|
|
| DIGORA |
| quote: | Originally posted by Tom2
I am trying to figure out your circlotron circuit.
This schematic is just and idea.
It is based on the Borgonio patent.
Please excuse me if my references to intellectual property are incomplete. |
hello
It can work in class AB ? |
|
|
| Tom2 |
| quote: | DIGORA said:
It can work in class AB ?
|
Yes.
The patent(4229706) refers to the diodes-resistors
(D1 R6 D2 R7, my schematic) connections as to help operate the circuit in class B.
I assume this can apply to class AB operation also.
Tom |
|
|
| EUVL |
To continue the discussion of a F5 like current feedback circlotron :
http://www.diyaudio.com/forums/show...534#post1549534
This would have been the circuit I would have proposed. I don't think it matters whether we use 2SK170 or 2SK369. I do think the J111 cascode is necessarily.
But I am also puzzled by Klaus's simulation results. Need to think first.
Patrick |
|
|
| KSTR |
Patrick,
Yes, I used pefectly balanced input (which is also quite important, of course).
Here Vgs at J2 (green) and Voltage accross R6 (blue). Not very nice at all....
Hhm, you might want to install LTSpice yourself, this makes evaluation much easier (models used for the transistors are found here on this forum, the JFETs were posted by syn08 and the MOSFETs by andy_c).
- Klaus |
|
|
| mrothacher |
| I have built and tested such a circuit and I'm working on the article now. I will find an appropriate place to publish once finished. |
|
|
| EUVL |
Did you experience the problem Klaus described, and if yes did you find a cure ?
Patrick |
|
|
| mrothacher |
| I haven't seen that problem, but our circuits aren't exactly the same. |
|
|
| juma |
| quote: | Originally posted by EUVL
...
This would have been the circuit I would have proposed. I don't think it matters whether we use 2SK170 or 2SK369. I do think the J111 cascode is necessarily.
...
Patrick |
Patrick's schematic as .gif file: |
|
|
| EUVL |
Thanks for converting to gif.
I just discover a mistake. The batteries should be the other way round.
Sorry guys.
Patrick
PS I believe I might have an explanation for the wild common mode bias swing, which is (mainly) at twice the frequency of the signal, but I do not have a cure yet. Need more thinking.
:scratch: |
|
|
| KSTR |
Seems that I found a remedy, maybe not the most elegant but it's working:
Buffering the gate drive with the right type of buffer (I missed this in my first try). This buffering also would help thermal compensation and is compatible to the current limiting (mabye a base reverse bias protection diode is needed). Freq. compensation seems necessary as the buffer unloads the resistors, I tried here this time with slightly inductive source resistors and an output snubber. It appears to be stable with capacitive loads in the nanofarads range and no resistive load (the situation when a cable is attached but not the speaker itself).
Now, at 20kHz, that common mode modulation is almost gone (reduced by the buffer's current gain) and the circuit behaves well and has good perfomance. Still it is not very intuitive what the root cause was, originally. At any rate, both capacitive gate currents got routed somehow through both resistors because the return points of these currents were on the "wrong side" of the load.
Matching of R7 to R12 is very critical... or a point where any mismatch introduces even order distortion if that is wanted. Other parameters seem less critical wrt matching.
Last not least, I'm very curious to see mrothacher's circuit....
- Klaus |
|
|
| EUVL |
Klaus,
If my head is functioning properly today, then your first circuit should work without this common mode current problem :
http://www.diyaudio.com/forums/atta...tamp=1214344739
The only change I would perhaps make is to refer R12,13 (100k) to Gnd, and cascode the JFETs. It is a bit difficult to explain, but it is to do with even harmonics of the input JFETs when driven by a differential signal, which would result in common mode currents flowing to Gnd, whereas the odd harmonic content of this current will cancel each other as they are of opposite polarity.
In the circuit by Circlomanen e.g., this JFET common mode current (at say 2x signal frequency) flows through the 22R resistor (& the 2200u cap) towards ground, then through the 100R output load resistor of the opposite leg, and then back to the power supply. This in turn creates a negative common-mode voltage at both outputs which is fed back to the JFET gate via the feedback resistor, ending up as positive feedback.
The solution would then be to connect the 22R resistors to each other (without the caps) and lift them from Gnd. This way, the JFET current is forced to flow through the feedback resistors and the load resistors before completing the loop to the power supply, and thus any positive common mode current will lead to an increase in JFET source voltage (i.e. returning back to negative feedback).
There should still be a very small common mode current at 2x signal frequency, but it should not be more than a few mA.
I don't have DSL at home, so I am not downloading LT Spice this weekend to prove my hypothesis. But I think this matches what you showed in the JFET and MOSFET currents.
Attached my new proposal.
Patrick |
|
|
| KSTR |
Hi Patrick,
I see you went through the same mind-boggling "in-the-head evaluation" horrors... :D
To cut things short, I found that the major contributor to the effect is the high and varying Cgd of the MOSFETs, that is, when I disable Cgd in the MOSFET models, et voíla, all is fine. The JFETs contribute to a far lesser amount. I virtually went through dozens of different hookups (including cascoding etc) but the most effective way to block the capacitive current flow in the JFET drain resistors is to buffer them with a follower in the way I've shown. The way feedback is arranged seems not to be important to the effect, meaning that your last proposal (which was my initial one) doesn't fix the problem, according to my findings.
How important this issue is in real life is still another matter since we don't expect a 10kHz or 20kHz full scale signal anyway (the tweeters will thank it). OTOH those two followers are an easy fix...
Finally, compared to the F5 the performance doesn't seem to be any superior, only added complexity (not least the supplies). The F5 is really hard to beat and that's why Admiral Nelson is the Master.
I like this quote of him very much:
"Like the products of other young designers, my first commercial product had everything but the kitchen sink in it. Now I strive to be like Picasso, who could draw a woman with a single pencil stroke and create a masterpiece."
- Klaus |
|
|
|
|
