GRollins said:
Yes, balanced negative feedback is being used here.
Earlier, I should have given the cross-coupling resistor a higher value commensurate with the change to 1.2k bias resistors. I've attached a new drawing showing all of these changes. Please note that this is still experimental, and all values are ballpark. I hope this makes things a little easier to follow.
Attachments
"No, the device does not know that you are using it as a
follower (CD) as opposed to voltage gain (CS). It has the
same characteristic in both cases."
I stand corrected.
Using a triode in a follower vs a gain stage sounds quite different, of course it cannot be a direct equivalent. For some reason I thought this would translate over to the depletion mode FETs.
"You mentioned JFETs, but the links you posted are both for MOSFETs. If you know of another source for power JFETs, please let us know."
I linked to them because they are both depletion mode types. The Supertex in particular has the same shape transfer curve as the Lovoltech part. The IXYS part will bias like the Lovoltech part, I have some samples to play with. I'm trying to get some TO-220 Supertex, the TO-92 ones are available for less than a buck from Mouser.
I have a 2KVA 40-40-40-40V toroid and four 110,000µF caps for the supply.
follower (CD) as opposed to voltage gain (CS). It has the
same characteristic in both cases."
I stand corrected.
Using a triode in a follower vs a gain stage sounds quite different, of course it cannot be a direct equivalent. For some reason I thought this would translate over to the depletion mode FETs.
"You mentioned JFETs, but the links you posted are both for MOSFETs. If you know of another source for power JFETs, please let us know."
I linked to them because they are both depletion mode types. The Supertex in particular has the same shape transfer curve as the Lovoltech part. The IXYS part will bias like the Lovoltech part, I have some samples to play with. I'm trying to get some TO-220 Supertex, the TO-92 ones are available for less than a buck from Mouser.
I have a 2KVA 40-40-40-40V toroid and four 110,000µF caps for the supply.
Grey,
I looked again at Nelson's ZV9 article this morning, and I do note an important difference. He modulates the cascode solely using the voltage across FET source resistor, which varies directly with output current. The circuit I propose includes this mechanism, but has the addition of FET drain modulation as a function of output voltage. This may trade off some distortion reduction for improved bilateral control - which could be just the ticket, given the strong and weak points of the circlotron.
If not, simply remove the cross-coupling resistor altogether, and you will have ZV9-style cascode modulation independently on each side of the circuit.
I looked again at Nelson's ZV9 article this morning, and I do note an important difference. He modulates the cascode solely using the voltage across FET source resistor, which varies directly with output current. The circuit I propose includes this mechanism, but has the addition of FET drain modulation as a function of output voltage. This may trade off some distortion reduction for improved bilateral control - which could be just the ticket, given the strong and weak points of the circlotron.
If not, simply remove the cross-coupling resistor altogether, and you will have ZV9-style cascode modulation independently on each side of the circuit.
I'm on the run, so this will have to be quick.
Joe,
I believe that negative feedback would be more appropriately applied to the Gates of the Lovoltech JFETs; two resistors, one from each output line to the opposite side's Lovoltech Gate, thus forming a resistive divider with the Gate's grounding resistor.
Give your circuit a try and let me know how things go. I'm open to being convinced.
djk,
The fact that the Lovoltech power JFETs are depletion mode isn't really relevant, although it makes biasing them somewhat more convenient.
TO-92 cases and 40mA bias are not going to work well in a power amp output stage. By the time you parallel enough devices to get up to a single amp of quiescent bias, you're going to have some ferocious problems with capacitance, etc. For comparison, each JFET/MOSFET cascode stack in the New-Tron runs at an amp, assuming that you build the circuit as posted. You're going to have to use 25 cascode stacks in parallel (at 40mA) to equal even one of those; 50 if you build a Circlotron. And that's not counting the second stack, or the fact that you could easily run the Lovoltechs at a higher bias current.
There's a more subtle problem, in that if you try to use the MOSFET devices you posted in a power amp output stage the current draw will vary widely as the amp drives the speaker. The current will swing well beyond the knee on the chart and you'll get some grossly non-linear behavior as a result. You'd be better off using the big TO-247 MOSFET alone and biasing it heavily enough that it's further over to the right on the chart.
It really isn't practical to try to coerce a MOSFET into behaving like a triode for this application. That's why the Lovoltech JFETs are so attractive. They can handle healthy current swings and stay linear. With the cascode keeping the voltage and heat off their backs, they do quite well indeed.
Grey
Joe,
I believe that negative feedback would be more appropriately applied to the Gates of the Lovoltech JFETs; two resistors, one from each output line to the opposite side's Lovoltech Gate, thus forming a resistive divider with the Gate's grounding resistor.
Give your circuit a try and let me know how things go. I'm open to being convinced.
djk,
The fact that the Lovoltech power JFETs are depletion mode isn't really relevant, although it makes biasing them somewhat more convenient.
TO-92 cases and 40mA bias are not going to work well in a power amp output stage. By the time you parallel enough devices to get up to a single amp of quiescent bias, you're going to have some ferocious problems with capacitance, etc. For comparison, each JFET/MOSFET cascode stack in the New-Tron runs at an amp, assuming that you build the circuit as posted. You're going to have to use 25 cascode stacks in parallel (at 40mA) to equal even one of those; 50 if you build a Circlotron. And that's not counting the second stack, or the fact that you could easily run the Lovoltechs at a higher bias current.
There's a more subtle problem, in that if you try to use the MOSFET devices you posted in a power amp output stage the current draw will vary widely as the amp drives the speaker. The current will swing well beyond the knee on the chart and you'll get some grossly non-linear behavior as a result. You'd be better off using the big TO-247 MOSFET alone and biasing it heavily enough that it's further over to the right on the chart.
It really isn't practical to try to coerce a MOSFET into behaving like a triode for this application. That's why the Lovoltech JFETs are so attractive. They can handle healthy current swings and stay linear. With the cascode keeping the voltage and heat off their backs, they do quite well indeed.
Grey
GRollins said:
I've used that kind of approach with MOSFETs and it works well. The method I show above would only work with FETs in "triode" mode, where their gain is a strong function of Vds. It might also be adapted for use with vacuum tubes, but that's another story.
I don't currently have a power FET circlotron to play with, but if I get around to trying this out before you, I'll let you know.
What a pity that there appears to be so little interest in this exciting concept from Grey !!
Although the circlotron is nothing new, the use of power JFETs in combination is certainly interesting to say the least. Recently, the Thorens TEM 3200, which is a circlotron based on tube frontend and MOSFET driver, has received top rating in Stereoplay (German HiFi magazine), above McIntosh MC501AC, Pass XA160 and Pass X600.
I certainly hope that Grey will keep us informed of progress.
Patrick
PS Link to the Thorens review :
http://www.stereoplay.de/sixcms/media.php/188/stp0106Thorens.pdf
Although the circlotron is nothing new, the use of power JFETs in combination is certainly interesting to say the least. Recently, the Thorens TEM 3200, which is a circlotron based on tube frontend and MOSFET driver, has received top rating in Stereoplay (German HiFi magazine), above McIntosh MC501AC, Pass XA160 and Pass X600.
I certainly hope that Grey will keep us informed of progress.
Patrick
PS Link to the Thorens review :
http://www.stereoplay.de/sixcms/media.php/188/stp0106Thorens.pdf
Had to get the oldest young 'un graduated from college (summa cum laude, honors, 4.0 average...go Katie!). Next we pack her off to Berkeley for graduate work in chemistry and biochemistry.
I've got a front end or two that I'll be posting shortly and I have scads of output stages, most of them Circlotrons, but some of other designs.
Grey
I've got a front end or two that I'll be posting shortly and I have scads of output stages, most of them Circlotrons, but some of other designs.
Grey
Grey,
How would you like this one (or a Power JFET version of) ?
http://www.diyaudio.com/forums/showthread.php?s=&threadid=55361
Patrick
(The original schematics in jpg is now too large to post !!!)
How would you like this one (or a Power JFET version of) ?
http://www.diyaudio.com/forums/showthread.php?s=&threadid=55361
Patrick
(The original schematics in jpg is now too large to post !!!)
NewT-FE2
Crossed fingers...if nothing goes wrong, I'm going to try to squeeze in a new front end schematic while things are quiet.
This one is based on the 2SK389. If you don't have those or can't get them, it's a curable problem. The Toshiba parts are still kinda sorta available, but there is a second source for them now and if there's enough interest, perhaps we can do a buy on them to save money. Every time I look at the Toshiba parts, the price has risen again--supply (out of production, inventory falling) and demand (from what I can tell, demand is more or less stable). The newer parts are in a different case and are expensive, but perhaps a buy would help with the price. I don't care about the case as long as I know ahead of time what I'm dealing with so I can get the PCB artwork to come out right.
Failing that, I have run the circuit with J310s, which are both available and affordable. The gain is lower and performance suffers a bit, but it works just fine. Other JFETs should work well, too.
This circuit is a little more elaborate than NewT-FE1. I've got a cascoded JFET front end feeding a MOSFET follower output. The outputs are biased hard, so use heatsinks on the MOSFETs or cut back on the bias a bit. The rails are asymmetrical at +45 and -10V. If you want to run +-45V, that's fine. The CCS can take it.
EUVL,
That's a pretty straight forward Circlotron. As such, it looks workable to me. I classify "X" feedback into two categories: Same Side Feedback (SSF), and Opposite Side Feedback (OSF). The Aleph-X is an example of OSF. The NewT-FE1 and NewT-FE2 front ends fall into the SSF category. I have other front end circuits which are OSF.
Incidentally, with minimal changes, the NewT-FE2 makes a pretty decent line stage.
Grey
Crossed fingers...if nothing goes wrong, I'm going to try to squeeze in a new front end schematic while things are quiet.
This one is based on the 2SK389. If you don't have those or can't get them, it's a curable problem. The Toshiba parts are still kinda sorta available, but there is a second source for them now and if there's enough interest, perhaps we can do a buy on them to save money. Every time I look at the Toshiba parts, the price has risen again--supply (out of production, inventory falling) and demand (from what I can tell, demand is more or less stable). The newer parts are in a different case and are expensive, but perhaps a buy would help with the price. I don't care about the case as long as I know ahead of time what I'm dealing with so I can get the PCB artwork to come out right.
Failing that, I have run the circuit with J310s, which are both available and affordable. The gain is lower and performance suffers a bit, but it works just fine. Other JFETs should work well, too.
This circuit is a little more elaborate than NewT-FE1. I've got a cascoded JFET front end feeding a MOSFET follower output. The outputs are biased hard, so use heatsinks on the MOSFETs or cut back on the bias a bit. The rails are asymmetrical at +45 and -10V. If you want to run +-45V, that's fine. The CCS can take it.
EUVL,
That's a pretty straight forward Circlotron. As such, it looks workable to me. I classify "X" feedback into two categories: Same Side Feedback (SSF), and Opposite Side Feedback (OSF). The Aleph-X is an example of OSF. The NewT-FE1 and NewT-FE2 front ends fall into the SSF category. I have other front end circuits which are OSF.
Incidentally, with minimal changes, the NewT-FE2 makes a pretty decent line stage.
Grey
Attachments
> Other JFETs should work well, too.
I would have just used 2 matched 2SK170s. Plenty around.
LSK389 is not widely available, expensive, and not really better than the Toshibas, as they claim.
> That's a pretty straight forward Circlotron.
I thought that is the Pass philosophy ? Certainly cannot be simpler than what he has drawn. What are the srawbacks of that circuit ?
Patrick
I would have just used 2 matched 2SK170s. Plenty around.
LSK389 is not widely available, expensive, and not really better than the Toshibas, as they claim.
> That's a pretty straight forward Circlotron.
I thought that is the Pass philosophy ? Certainly cannot be simpler than what he has drawn. What are the srawbacks of that circuit ?
Patrick
The Pass Philosophy??? Crrently I have a couple prototype line amps that you see below. I think it deffinatly has it's roots in the Pass Philosophy... It is my first answer to the "How to drive the New-Tron" question... I've also been thinking in bridged circles for a while and the topology below will do the job. It simulates pretty good and an unmatched component version measures about 10X the sim data so I'm anctous to measure the second "matched" prototype...
Comments
Comments
Attachments
EUVL,
Here in the US, the 2SK170 is not readily available. If you are able to find them, you have no control over whether you're getting GR, BL, or VI. Actually, strike that--I've never even seen a 2SK170 VI. Like the Nonesuch, I've come to doubt that they even exist. The LSK170 is available and you can order the Idss you want. The price is somewhat higher than what we pay for the Toshiba parts, but that may improve if bought in quantity.
Note that although the '170 is available, you still have to buy in quantity and hand match them to simulate a '389. Your cost skyrockets accordingly.
Toshiba parts are not--unfortunately--the kind of thing you can just walk into a store and buy here in the US. They're all black market.
Speaking of which, I'd love to try the Toshiba 2SK2013/2SJ313 (I think I've got those numbers right), but they do not seem to be available anywhere at any price here in the US. If anyone can suggest a source, I'd appreciate it. Yes, I know Borbely lists them, but he has a steep minimum order and I have a feeling that the parts themselves can be had more cheaply...if only I can find the damned things. I am reluctant to sink a lot of money into those parts until I can try them in a circuit I've got going. If they don't work, I don't want to be sitting on a bunch of the silly things.
The circuit referenced earlier is a standard Circlotron. It's essentially the same circuit I've posted, just drawn differently.
Most tube circuits were pretty simple. Solid state adaptations of those topologies tend to remain pretty simple unless someone gets a wild hair and decides to "improve" the circuit. Some improvements work. Most don't. To the extent that it's a simple circuit, I'd say that it's certainly in line with Nelson's KISS circuits.
flg,
The circuit looks good, but I think it needs a couple of power JFETs.
(ahem...)
Grey
Here in the US, the 2SK170 is not readily available. If you are able to find them, you have no control over whether you're getting GR, BL, or VI. Actually, strike that--I've never even seen a 2SK170 VI. Like the Nonesuch, I've come to doubt that they even exist. The LSK170 is available and you can order the Idss you want. The price is somewhat higher than what we pay for the Toshiba parts, but that may improve if bought in quantity.
Note that although the '170 is available, you still have to buy in quantity and hand match them to simulate a '389. Your cost skyrockets accordingly.
Toshiba parts are not--unfortunately--the kind of thing you can just walk into a store and buy here in the US. They're all black market.
Speaking of which, I'd love to try the Toshiba 2SK2013/2SJ313 (I think I've got those numbers right), but they do not seem to be available anywhere at any price here in the US. If anyone can suggest a source, I'd appreciate it. Yes, I know Borbely lists them, but he has a steep minimum order and I have a feeling that the parts themselves can be had more cheaply...if only I can find the damned things. I am reluctant to sink a lot of money into those parts until I can try them in a circuit I've got going. If they don't work, I don't want to be sitting on a bunch of the silly things.
The circuit referenced earlier is a standard Circlotron. It's essentially the same circuit I've posted, just drawn differently.
Most tube circuits were pretty simple. Solid state adaptations of those topologies tend to remain pretty simple unless someone gets a wild hair and decides to "improve" the circuit. Some improvements work. Most don't. To the extent that it's a simple circuit, I'd say that it's certainly in line with Nelson's KISS circuits.
flg,
The circuit looks good, but I think it needs a couple of power JFETs.
(ahem...)
Grey
UHHmmm, Yes, your right. Something like that is in order. I wouldn't want to stretch the thread to far...But, I want to use my JFETs in a variation combining the ZV7T and ZV9. I guess I'm free to call it ZV9T. There has been a little chat about that path already. I beleive there is something working out there that is at least 1/2 of that... I have this real free time crunch going on too... I'm being Dev#$%ed and I have to sell my house etc etc etc... So, I don't have much left thats not in a box but the simulator and the kitchen table... I would also like to develope a good Pspice model for the JFET while I'm winding down things at the homestead... I'm workin on it...
Grey,
You can get 2SK170BLs AND Vs from Fibra Brandt Semiconductors (just google) at reasonable price, and the specific type that you want. I actually have more problems getting GR. I have bought enough Vs from him, so they definitely exist.
I have located a possible source in Japan for 2SK2013/2SJ313. They are not expensive, not more than any other Toshiba TO220 MOSFETs. If you are interested, please contact me by private mail. They are also not available in Europe, or anywhere else other than Japan. I actually asked Toshiba here.
Patrick
You can get 2SK170BLs AND Vs from Fibra Brandt Semiconductors (just google) at reasonable price, and the specific type that you want. I actually have more problems getting GR. I have bought enough Vs from him, so they definitely exist.
I have located a possible source in Japan for 2SK2013/2SJ313. They are not expensive, not more than any other Toshiba TO220 MOSFETs. If you are interested, please contact me by private mail. They are also not available in Europe, or anywhere else other than Japan. I actually asked Toshiba here.
Patrick
flg,
Just teasing, of course. The circuit will work without power JFETs...just not as well.
(Who, me? I'm not nudging people to use these things...)
EUVL,
I haven't checked the Euro/dollar conversion recently, but the prices Fibra-Brandt quotes seem high to me. Especially with the US dollar falling against other currencies.
Regarding the 2SK2013/2SJ313: Why on Earth would a manufacturer make a product, then make such an effort not to sell it? Mind boggling. You'd think that they would make at least a rudimentary effort to distribute the product so they could sell a few.
Grumble.
Alain,
I had problems with consistency with NTE product--so much so that I quit buying anything they sell. I would buy one or two devices, do some R&D, then go back and buy more, only to discover the the next ones didn't behave the way the first ones did. I've also had problems with parts that weren't good matches for the ones they were supposedly replacing, etc. And the price was usually high. I gave up.
The worst case was a high voltage regulator I built for a tube circuit. It used a MOSFET as a pass device--I don't remember which NTE part number, this was ten years ago. I built three regulators with NTE MOSFETs I had on hand. All worked perfectly. I bought more and they oscillated horribly. Nothing would make them behave. I eventually had to change the regulator circuit. Boy, was I peeved!
Grey
Just teasing, of course. The circuit will work without power JFETs...just not as well.
(Who, me? I'm not nudging people to use these things...)
EUVL,
I haven't checked the Euro/dollar conversion recently, but the prices Fibra-Brandt quotes seem high to me. Especially with the US dollar falling against other currencies.
Regarding the 2SK2013/2SJ313: Why on Earth would a manufacturer make a product, then make such an effort not to sell it? Mind boggling. You'd think that they would make at least a rudimentary effort to distribute the product so they could sell a few.
Grumble.
Alain,
I had problems with consistency with NTE product--so much so that I quit buying anything they sell. I would buy one or two devices, do some R&D, then go back and buy more, only to discover the the next ones didn't behave the way the first ones did. I've also had problems with parts that weren't good matches for the ones they were supposedly replacing, etc. And the price was usually high. I gave up.
The worst case was a high voltage regulator I built for a tube circuit. It used a MOSFET as a pass device--I don't remember which NTE part number, this was ten years ago. I built three regulators with NTE MOSFETs I had on hand. All worked perfectly. I bought more and they oscillated horribly. Nothing would make them behave. I eventually had to change the regulator circuit. Boy, was I peeved!
Grey
Yep, I realize that Grey. but I was already doing Circlotron before the Z9, maybe even before Z8, and before that I was finishing up the "LH". I wasn't willing to complicate things with changing a good looking output device and adding a cascode that may not be necessary???
I really like the Inductor load idea of ZV7 though, especialy with a "triode like" device. I may go on to do the follower thing with the JFETs later... That will be down the road a bit.
How about this for a good reason, speaking of down the road. My Mom & Dad are getting old and I've been in the Phx area since I was 18. Next Sunday I'll be 48. After my Dev$%^*ce, I' want to go live with them back in R.I. and help them out and try to move on myself... My dad was in electronics before me and he's a Ham... I will start the tube like thinking when I can convince him to join me on a project. I've really wanted to try to adapt the Ultra-Linear tube Topology to a cascoded JFET circuit, any ideas on that??? We may actually have been discussing it for months now???
This also means, in a month or so I will dissapear from the forum while me and my doggies drive across the USA in search of greener pastures??? or something... Maybe I'll stop and say hello Grey
It did not seem like I was contributing there...Forgive my mood... O.K. If anyone cares, I got my last 74s &170s from the AmpsLab Guy. Singapore I think??? He has a few other worthwhile components too. Do a dogpile (google for you people without dogs) on ampslab. I wouldn't call it a deal but, I have some...
I work at ON Semi and we don't have any good complementary JFETs either. Do We???
I really like the Inductor load idea of ZV7 though, especialy with a "triode like" device. I may go on to do the follower thing with the JFETs later... That will be down the road a bit.
How about this for a good reason, speaking of down the road. My Mom & Dad are getting old and I've been in the Phx area since I was 18. Next Sunday I'll be 48. After my Dev$%^*ce, I' want to go live with them back in R.I. and help them out and try to move on myself... My dad was in electronics before me and he's a Ham... I will start the tube like thinking when I can convince him to join me on a project. I've really wanted to try to adapt the Ultra-Linear tube Topology to a cascoded JFET circuit, any ideas on that??? We may actually have been discussing it for months now???
This also means, in a month or so I will dissapear from the forum while me and my doggies drive across the USA in search of greener pastures??? or something... Maybe I'll stop and say hello Grey
It did not seem like I was contributing there...Forgive my mood... O.K. If anyone cares, I got my last 74s &170s from the AmpsLab Guy. Singapore I think??? He has a few other worthwhile components too. Do a dogpile (google for you people without dogs) on ampslab. I wouldn't call it a deal but, I have some...
I work at ON Semi and we don't have any good complementary JFETs either. Do We???
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.