George,
How much voltage swing were you able to get out of a 6SN7 at reasonable distortion?
I have a dream of making a 6L6 AB2 triode unity-coupled amp and need about 580Vpk-pk drive. I will have 1000V B+ available for the driver, but I have no idea what the 6SN7 curves look like above 500V. If the tube remains fairly linear up there, I could use two differential stages like you are. Otherwise, I will probably have to use the trick I mentioned earlier in this thread and bias the stage following the LTP at a low plate voltage and let it clip at Vg=0. This would obviously rule out a differential stage for that stage so I would have to go common cathode. It would suffer from some amount of imbalance due to the inevitable differences in gain between the two sections of the 6SN7. Might not be too bad, though. I'll just have to build it and see.
How much voltage swing were you able to get out of a 6SN7 at reasonable distortion?
I have a dream of making a 6L6 AB2 triode unity-coupled amp and need about 580Vpk-pk drive. I will have 1000V B+ available for the driver, but I have no idea what the 6SN7 curves look like above 500V. If the tube remains fairly linear up there, I could use two differential stages like you are. Otherwise, I will probably have to use the trick I mentioned earlier in this thread and bias the stage following the LTP at a low plate voltage and let it clip at Vg=0. This would obviously rule out a differential stage for that stage so I would have to go common cathode. It would suffer from some amount of imbalance due to the inevitable differences in gain between the two sections of the 6SN7. Might not be too bad, though. I'll just have to build it and see.
I am an early adopter of the Tubelab SE amp that was running on a pair of Decware HDT's I built...OUTSTANDING amp. Unfortunately while working on a Passlab SS amp a small clipped lead managed to careen off 2 walls and land on one of the CCS's blowing a couple of resistors and the CCS. This misfortune has turned into an opportunity to build the amp into a steel chassis with point to point wiring and the customizations for a dedicated 300B amp...still in progress.
But what I have wanted since first seeing "The Beast" was a PP 300B amp and it looks like Tubelab may yet present a schematic for such (the not so simple PP amp
Looking forward to it!!! This forum is truly a great tool. I'll shut up now and go back to lurking.
But what I have wanted since first seeing "The Beast" was a PP 300B amp and it looks like Tubelab may yet present a schematic for such (the not so simple PP amp
Looking forward to it!!! This forum is truly a great tool. I'll shut up now and go back to lurking.
YES, they are called pentodes. I read this on Gary Pimms web site. I tried it and found pretty much what he reported. You can get a bunch of gain, low distortion, and high voltage swing. The drawback, you have to tweak each individual tube for lowest distortion, and it does drift a bit as the tube has been on for a few hours. Some tubes respond better than others. Gary reports poor results with the 6CL6. I thought it was OK. The 12BY7 can put out a bunch of voltage, but tend to vary more than others from tube to tube.
http://www.pacifier.com/~gpimm/drivers.htm
Well, that depends on how you interpret the "specs" and how much power supply you have.
OK, not that much! Seriously I was running 500 - 550 volts with a resistive load and getting up to 400+ Vp-p. I really didn't do to much testing, or need much more drive. I did find that not all 6SN7's are created equal when operated outside the "normal" applications. I have a box full of 6SN7's I simply plugged a bunch into the tag board and selected a pair that cranked out a bunch of voltage. The little short guy with the shiny chrome head worked the best as the first stage. The tall guy seemed to have no problem with making gobs of drive voltage at low distortion. I never measured the actual distortion of the driver stage, just the total test amp. The driver output was observed with a scope probe on each plate. It would seem that some tubes go nonlinear at the plate attempts to pull down below 100 volts, some go to 50 without an issue. At the other extreme, the maximum DC plate voltage spec for some species of 6SN7 is 450 volts. The 6SN7 was specified for use as a TV vertical output tube. This means that it can handle an inductive load (vertical output transformer) implying that a peak plate voltage can be expected to reach twice the DC voltage. It is in fact specified for a 1500 volt peak plate voltage. I have not found the "limit" yet, but I don't have a 1000 volt supply on my driver.......yet.
Yes, I need a new "Beast" or at least a new power supply for my old one. The reality is that the old Beast played flawlessly for the first few years of its life, and has developed a crackling sound that gets worse as it gets hot. Attemps to fix it have failed.
I am indeed working on a new driver board. The schematics posted earlier were mostly for academic purposes, but a working amplifier is coming. The time frame is highly uncertain at this time. I spent this evening helping a friend clean out his office and pack 18 years of his life into his car. This is happening weekly where I work.
That's good news. Like I said earlier, I'm planning on at least initially testing the last voltage amplifier as a grounded cathode amp with 10M90S cascode. I will have approximately 750V B+ for this stage as the current scribblings indicate. I'm just a little worried that those curves might start bunching up at higher voltages. If only I had a curve tracer.
I plan on doing a +/-500V supply with a Hammond 278CX. Maidas would be nice. I've never seen anyone do a -500V Maida. Got to find a pass device...
The Maida schematic shows two NPN transistors in darlington configuration as the pass device. I have never built one this way. I just use a mosfet. Some component value tweaking is required. The zener voltage must be raised because of the gate voltage requirement of a mosfet and the actual diode voltage depends on the individual fet. I think I used a 15 volt zener. I used my old favorite the 2SK2700 for most of my experiments. I currently have a "Mega Maida" operating that uses a STW5NK100Z mosfet from ST micro. That fet is good for several amps at 1000 volts, I haven't blown one yet. In theory the pass device only sees the input / output differential of the regulator, but I needed some big fets for hybrid amp experiments so I got the ST guys, and some big heat sinks.
A negative regulator can be made with a negative adjustable regulator like the LM337 and a P channel fet. The only P cannel fet that I have much experience with is the Fairchild FQP1P50 which is rated to 500 volts and 1 amp. I was using it in some hybrid driver experiments where low capacitance was needed. I have not tried to make a negative Maida yet though.
I want one too. I think that the hardware would be fairly easy. The Windows programming for the display and the USB interface is the show stopper for me though. It is possible to plot useful curves with only two power supplies a voltmeter, a current meter, and Excel. I have actually done this.
I use the FQPF1P50 all the time. The plastic case is very convenient for chassis mounting. I know that the pass device shouldn't see the full voltage, but still it makes me nervous and I wish there were an FQPF1P80 or something. In all of the schematics I have ever seen, the pass device is rated for more than the raw DC voltage, about 580V in my case.
Why not just copy the data over to the PC and just write a .csv file? Then you could plot it wherever you like, Excel, Matlab, etc. Easy software.
I will eventually build one. Something with a low duty cycle that can draw curves well outside the max dissipation area. I hate how most data sheet curves end just when it starts getting interesting...
Why not just copy the data over to the PC and just write a .csv file? Then you could plot it wherever you like, Excel, Matlab, etc. Easy software.
I will eventually build one. Something with a low duty cycle that can draw curves well outside the max dissipation area. I hate how most data sheet curves end just when it starts getting interesting...
I haven't had any time to finish the driver board since all of my free time this year has been spent sorting tubes. About 15,000 tubes have gone through my hands in the past 4 weekends, and I am not even halfway done. I will probably have room to keep a few thousand. Decisions, decisions, decisions........ I am going to need a curve tracer.
I had visions of some PIC chip controlled vacuum tube torture chamber connected to a PC via USB. It should be capable of real time display with a couple of knobs that let you control the meltdown while you watch, like the old TEK stuff. I can handle the hardware design and the PIC chip programming, but the USB and Windows stuff confuses me. I have come to the realization that I am never going to build this, especially now that all of the software engineers that I used to hang out with at work have been laid off
I was beginning to think about using one of those single board computers found in Circuit Cellar magazine ads to control the torture chamber and dump the data on a SD card in csv format. The circuit on Jackinnj's web site is probably a realistic compromize. I don't have the time right now any way, but it will happen some day.
Yeah, low duty cycle pulses, and 1KV capability, thats the ticket. Of course variable duty cycle could be used to see how the curves change as the average dissipation is increased even though the voltage and current is not changing.
I had visions of some PIC chip controlled vacuum tube torture chamber connected to a PC via USB. It should be capable of real time display with a couple of knobs that let you control the meltdown while you watch, like the old TEK stuff. I can handle the hardware design and the PIC chip programming, but the USB and Windows stuff confuses me. I have come to the realization that I am never going to build this, especially now that all of the software engineers that I used to hang out with at work have been laid off
I was beginning to think about using one of those single board computers found in Circuit Cellar magazine ads to control the torture chamber and dump the data on a SD card in csv format. The circuit on Jackinnj's web site is probably a realistic compromize. I don't have the time right now any way, but it will happen some day.
Yeah, low duty cycle pulses, and 1KV capability, thats the ticket. Of course variable duty cycle could be used to see how the curves change as the average dissipation is increased even though the voltage and current is not changing.
These are excellent. I have used PIC but Rabbit is easier, cheaper, and code examples are better. Rabbit's IDE is much better than the CCS IDE for PIC and it comes with the dev kit, which is within a hobbyist's budget. I have used this dev kit for quick turnaround projects and programming for the mini-SD card is easy. I imagine ethernet wouldn't be too difficult either.
I got all the stuff to do AVR. But I am leaning toward RS232 with
standard SCPI commands. Gotta put all this Labview training to
practice some day soon before I forget how.
I'm still not sure why my company was in such a panic to get me
Labview certified??? Nothing ever came of it (yet)...
Most of the testing I do is non-repetitive small unique batches.
Far quicker to do manually than set up something automated.
But a high voltage curve tracer is a plausible project I could
probably squeeze in. Evenings or slow days. I'd have to have
all my regular work finished first.
standard SCPI commands. Gotta put all this Labview training to
practice some day soon before I forget how.
I'm still not sure why my company was in such a panic to get me
Labview certified??? Nothing ever came of it (yet)...
Most of the testing I do is non-repetitive small unique batches.
Far quicker to do manually than set up something automated.
But a high voltage curve tracer is a plausible project I could
probably squeeze in. Evenings or slow days. I'd have to have
all my regular work finished first.
That rabbit kit was $299. Cheap?? A little pricey for my taste.
Though I like the backward compatible Z80 instruction set.
Mikroe development kits and compilers are cheap and good.
Most having more features than you can shake a stick at.
Your choice: PIC/AVR/ARM , BASIC/C/PASCAL
http://www.mikroe.com/
Though I like the backward compatible Z80 instruction set.
Mikroe development kits and compilers are cheap and good.
Most having more features than you can shake a stick at.
Your choice: PIC/AVR/ARM , BASIC/C/PASCAL
http://www.mikroe.com/
I got all the stuff to do PIC. The IDE (MpLAB) is free from their web site, so is a limited "C" complier, and the In Circuit Debugger / programmer is reasonable. Microchip used to come into the plant where I work and give tutorial classes. Development boards, full version compilers, and ICD's were handed out in those classes. I have been using PIC chips for about 15 years, so they are my default choice. There have been several articles, and book chapters written on interfacing a PIC to an SD card, so that is a possibility. I am open to another choice if it looks better. It must wait for now, though.
It don't matter much what chunk of sand does the dirty work.
Whatever feels familiar and gets the job done.
I do recommend you look into SCPI "Skippy" commands as a
potential protocol for talking to the darn thing over RS232.
Already well standardized. Same command structure as GPIB.
I recall seeing a fairly complete Wiki on the subject.
Labview and Matlab already talk to SCPI without having to
completely write new virtual instruments to collect the data.
Certainly a shortcut toward putting together a GUI.
Whatever feels familiar and gets the job done.
I do recommend you look into SCPI "Skippy" commands as a
potential protocol for talking to the darn thing over RS232.
Already well standardized. Same command structure as GPIB.
I recall seeing a fairly complete Wiki on the subject.
Labview and Matlab already talk to SCPI without having to
completely write new virtual instruments to collect the data.
Certainly a shortcut toward putting together a GUI.
I guess my beef isn't really with MicroChip. It's the CCS compiler that I (actually my employer)paid $500 for. My main problem with it was that it didn't work. I could not get it to execute code. I spent weeks on this problem only to find out that older versions of the compiler did not have this problem. I had to hack the hex file in order to get code to execute on the PIC. I contacted CCS and they said that I had to pay them more money in order to get updates to fix their product that never worked, not even once. This experience left a bad taste in my mouth, worse than that one time I got gasoline in my mouth.
Meanwhile, I had ordered a Rabbit development kit and got the product completed in a couple of days. It was quite satisfying to tell the CCS guys, "Never mind, I already got it working on a Rabbit."
The Rabbit core board comes with a lot of hardware (the ethernet and SD card reader) already done for you. You just have to do the motherboard with the power supply controls.
Do they come with the compiler/IDE?(the Rabbit does) That part was unclear to me as I was browsing the site. (It just says that it comes with 'software') If not, it would end up being quite a bit more than $300 for me to get set up with. I like the model with the big LCD display. The board is kind of big, though.
Agreed.
Meanwhile, I had ordered a Rabbit development kit and got the product completed in a couple of days. It was quite satisfying to tell the CCS guys, "Never mind, I already got it working on a Rabbit."
The Rabbit core board comes with a lot of hardware (the ethernet and SD card reader) already done for you. You just have to do the motherboard with the power supply controls.
Do they come with the compiler/IDE?(the Rabbit does) That part was unclear to me as I was browsing the site. (It just says that it comes with 'software') If not, it would end up being quite a bit more than $300 for me to get set up with. I like the model with the big LCD display. The board is kind of big, though.
Agreed.
Check out Arduino or compatibles. It's really caught on in a big way. Cheap too.
You can even use free gcc as a development tool.
http://www.freeduino.org/
http://www.ladyada.net/make/boarduino/
Futurelec has good deals on controllers. Arm or atmel are easy to use.
http://www.futurlec.com/BasicControllers.shtml
You can even use free gcc as a development tool.
http://www.freeduino.org/
http://www.ladyada.net/make/boarduino/
Futurelec has good deals on controllers. Arm or atmel are easy to use.
http://www.futurlec.com/BasicControllers.shtml
I tried the CCS compiler a few years back on the 16F series parts. It worked but I did have some problems. Microchip now has their own compiler (3 different versions) that appears to be GNU based for their 18F and higher series parts. There is a free version on their web site that doesn't do as much optimization after compilation as the full version. The full version is not cheap, but I got the 18F and 24F versions in a training class. These compilers seem to work quite good.
Agreed. There are all sorts of architectural differences between processors that make choosing the controller for say a cell phone a formidal task that requires a design commitee and months of study. For a simple project like this, it won't make any difference at all as long as the basics are covered (A/D, SPI, enough I/O, etc).
I never tried Rabbit, probably because of the $300. I already have two Microchip Explorer 16 boards and they didn't cost me anything, so that is the logical choice, but there is no SD card slot.
I have seen some Arduino campatible boards on Ebay for cheap, don't know if they are any good or not.
My beef with PIC is "W", and I don't just mean Bush...
One working register means two instructions to pass any
data or get anything done. And banked memory and I/O
mapping of the most absurdly inconvenient sort. Absolutely
wasteful of code space, cycles, and ultimately my patience...
Some of the new dsPIC I hear this is fixed. As PIC seems
to change their instruction set every year or two...But with
the old Rabbit(Z80) and AVR there have always been plenty
of true registers. The instruction set never been a problem.
You can get the full AVR Studio from Atmel, I don't think it
costs anything, but you do have to register. But thats just
for C and/or Assembler.
As I only know oldskool Basic, Assembler, and now Labview.
C might as well be ancient Sumerian as far as I can read it.
Assembler I just don't have the patience anymore.
I need a Basic compiler if I want anything done fast. And
Mikeroe has one of the better Basic compilers. FastAVR is
another. And Oshonsoft, though kinda light, its not light
in a bad way.
You just need to dig around Mikeroe's sight for the bundle
deals. You can get a complete kit (including both LCDs) with
the compiler of your choice for under $250. And socketed to
program as many cheap stand-alone controllers as you wish.
The only extra you might buy ala cart is Ethernet if you really
need one.
Which is $50 cheaper than the Rabbit with no peripherals
at all, but SD and Ethernet. And the chip ain't coming off.
Which is gonna cost you an arm and leg if you ever care
to duplicate your project.
Arduino is also AVR based, but they are off thier own trip.
I don't really understand whats the deal with them.
Good alternatives not yet mentioned: Parallax Propeller,
Coridium's ARM (with Basic interperetrer shell built in),
and the PICaxe if you don't mind it being a little slow....
One working register means two instructions to pass any
data or get anything done. And banked memory and I/O
mapping of the most absurdly inconvenient sort. Absolutely
wasteful of code space, cycles, and ultimately my patience...
Some of the new dsPIC I hear this is fixed. As PIC seems
to change their instruction set every year or two...But with
the old Rabbit(Z80) and AVR there have always been plenty
of true registers. The instruction set never been a problem.
You can get the full AVR Studio from Atmel, I don't think it
costs anything, but you do have to register. But thats just
for C and/or Assembler.
As I only know oldskool Basic, Assembler, and now Labview.
C might as well be ancient Sumerian as far as I can read it.
Assembler I just don't have the patience anymore.
I need a Basic compiler if I want anything done fast. And
Mikeroe has one of the better Basic compilers. FastAVR is
another. And Oshonsoft, though kinda light, its not light
in a bad way.
You just need to dig around Mikeroe's sight for the bundle
deals. You can get a complete kit (including both LCDs) with
the compiler of your choice for under $250. And socketed to
program as many cheap stand-alone controllers as you wish.
The only extra you might buy ala cart is Ethernet if you really
need one.
Which is $50 cheaper than the Rabbit with no peripherals
at all, but SD and Ethernet. And the chip ain't coming off.
Which is gonna cost you an arm and leg if you ever care
to duplicate your project.
Arduino is also AVR based, but they are off thier own trip.
I don't really understand whats the deal with them.
Good alternatives not yet mentioned: Parallax Propeller,
Coridium's ARM (with Basic interperetrer shell built in),
and the PICaxe if you don't mind it being a little slow....
Why bother with a microP controller at all. Most of the power supply manufacturers have produced power supplies with RS-232 remote control at some time. Used ones on Epay.
Xantrex has the 600V XHR and XFR series with RS-232 or GPIB options. Their XT series has 120V and 250V versions with RS-232 or GPIB options. All have 16 bit control and readback of V and I (the internal D/A and A/D chip accuracy).
The GPIB versions can be converted to RS-232 with addition of a chip, a connector and some bypass caps. (Oh, also the Eprom code needs reprogramming to the RS-232 version, which I can send the data file for, or program the chip on my Willem Eprom programmer card - something like $35 on Epay for the Willem card if I remember correctly. It also programs some of the uP controller chips too.)
Most of the other pwr supplies around, that I've seen specs for, have 12 bit control and readback, which should be adequate for curve tracing.
Need a Basic or Vis. Basic, etc. program to control them then.
Don
Xantrex has the 600V XHR and XFR series with RS-232 or GPIB options. Their XT series has 120V and 250V versions with RS-232 or GPIB options. All have 16 bit control and readback of V and I (the internal D/A and A/D chip accuracy).
The GPIB versions can be converted to RS-232 with addition of a chip, a connector and some bypass caps. (Oh, also the Eprom code needs reprogramming to the RS-232 version, which I can send the data file for, or program the chip on my Willem Eprom programmer card - something like $35 on Epay for the Willem card if I remember correctly. It also programs some of the uP controller chips too.)
Most of the other pwr supplies around, that I've seen specs for, have 12 bit control and readback, which should be adequate for curve tracing.
Need a Basic or Vis. Basic, etc. program to control them then.
Don
I have 5 power supplies already controlled by GPIB.
6,25,25,30,60. I don't know how tolerant they'd be
of daisy chaining to a higher voltage? They are all
wayyy too expensive to be screwing around with.
Tough if I had one I owned personally, talking to
it over the GPIB bus here wouldn't be an issue.
I've not seen the supply you refer to, much less
at that kinda price.
And to trace anything, you need to control at least
two parameters while measuring the third... One
programmable volatge alone ain't gonna cut it.
6,25,25,30,60. I don't know how tolerant they'd be
of daisy chaining to a higher voltage? They are all
wayyy too expensive to be screwing around with.
Tough if I had one I owned personally, talking to
it over the GPIB bus here wouldn't be an issue.
I've not seen the supply you refer to, much less
at that kinda price.
And to trace anything, you need to control at least
two parameters while measuring the third... One
programmable volatge alone ain't gonna cut it.
"I have 5 power supplies already controlled by GPIB.
6,25,25,30,60. I don't know how tolerant they'd be
of daisy chaining to a higher voltage? "
There should be a differential Vout to ground specification for the supplies. The Xantrex XHR is rated +/-600V from safety ground for example. 400V diffl. is probably more typical (Kepco), but I've seen some Lambda Zup units with a 0V rating! So watch out.
GPIB controlled power supplies are quite common on Epay, especially if you already are set up for that. (Kepco, Lambda, HP, Xantrex, Sorensen, ...)
RS-232 remote control supplies often get bid to Hell_o. That's why I figured out how to convert Xantrex GPIBs to RS-232s. XHR 600V units go for around $200 to $500 with a remote. But are sometimes much cheaper ($60) if not working. They appear regularly on Epay. XT units go for $50 to $150. But XT 120 V or 250 V units are quite scarce on Epay (like maybe years to find one). The remote card can be salvaged from a cheap low voltage unit though, so thats not an issue.
(most of mine were bargain ones that I knew I had to fix, but it's been surprising how often someone just couldn't set the DIP switches right on the back. Otherwise, its usually a power Mosfet switch that went, or the output current sense resistor is blown.)
Unfortunately, Xantrex does not supply service manuals for their units, but Sorensen has sold many Xantrex units under their name, and they have often supplied service manuals for them. Unfortunately, Xantrex bought Sorensen recently and so no more manuals except on Epay.
By the way, although the switching type supplies generally have a current limit control knob (and remote too if optioned), it only controls average current into the big output capacitor. So dead shorting one of these ( the XHR is a switcher for example) will blow the current sense resistor instantly unless its a huge thing.
Don
6,25,25,30,60. I don't know how tolerant they'd be
of daisy chaining to a higher voltage? "
There should be a differential Vout to ground specification for the supplies. The Xantrex XHR is rated +/-600V from safety ground for example. 400V diffl. is probably more typical (Kepco), but I've seen some Lambda Zup units with a 0V rating! So watch out.
GPIB controlled power supplies are quite common on Epay, especially if you already are set up for that. (Kepco, Lambda, HP, Xantrex, Sorensen, ...)
RS-232 remote control supplies often get bid to Hell_o. That's why I figured out how to convert Xantrex GPIBs to RS-232s. XHR 600V units go for around $200 to $500 with a remote. But are sometimes much cheaper ($60) if not working. They appear regularly on Epay. XT units go for $50 to $150. But XT 120 V or 250 V units are quite scarce on Epay (like maybe years to find one). The remote card can be salvaged from a cheap low voltage unit though, so thats not an issue.
(most of mine were bargain ones that I knew I had to fix, but it's been surprising how often someone just couldn't set the DIP switches right on the back. Otherwise, its usually a power Mosfet switch that went, or the output current sense resistor is blown.)
Unfortunately, Xantrex does not supply service manuals for their units, but Sorensen has sold many Xantrex units under their name, and they have often supplied service manuals for them. Unfortunately, Xantrex bought Sorensen recently and so no more manuals except on Epay.
By the way, although the switching type supplies generally have a current limit control knob (and remote too if optioned), it only controls average current into the big output capacitor. So dead shorting one of these ( the XHR is a switcher for example) will blow the current sense resistor instantly unless its a huge thing.
Don
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.