PAINT_KIT.JAR
> Hi Nick,
>
> Yeah getting paintcc to work is not trivial. You must follow the
> instructions on the site exactly in order for it to work. (you are
> actually running a script that invokes a java session, and then loads
> and runs paintcc) Make sure you have the latest version of sun java
> before attempting to run this program. Go to Sun website to get that
> if you need it.
>
> You also need to capture the tube curves you are attempting to match
> and save them to a gif file. I specifically recommend this program to
> do the capture: http://www.screenprint32.com
>
> You need to save the gif file before running paint_kip as you will
> call it in the script you run to start it.
>
> This curve matcher is not at all user friendly!
>
> There is a learning curve to all of this stuff. Just take your time
> and bang away at it..
Hi Kevin,
Regards www.geocities.com/dmitrynizh/tubeparams_image.htm: I'm glad
you recommended it to Nick and also offered some practical tips,
thanks. Usability is still suboptimal. At the bottom of the webpage, I
describe some possible improvemnts for the package. It looks as
allowing the jar to be invoked via "java web start" and providing
standard file dialog windows for input/output would greatly help those
lacking java command line experience. I'll try to improve that.
Regards screenprint32: I personally use LVPro to capture the screen
and save the result as a GIF file, but it is good that there is an alternative.
I found the link you posted currently timeouts, though. PC World Downloads still has
the installation at http://www.pcworld.com/downloads/file/fid,22745-order,1-page,1-c,alldownloads/description.html
-Dmitry
www.geocities.com/dmitrynizh
> Hi Nick,
>
> Yeah getting paintcc to work is not trivial. You must follow the
> instructions on the site exactly in order for it to work. (you are
> actually running a script that invokes a java session, and then loads
> and runs paintcc) Make sure you have the latest version of sun java
> before attempting to run this program. Go to Sun website to get that
> if you need it.
>
> You also need to capture the tube curves you are attempting to match
> and save them to a gif file. I specifically recommend this program to
> do the capture: http://www.screenprint32.com
>
> You need to save the gif file before running paint_kip as you will
> call it in the script you run to start it.
>
> This curve matcher is not at all user friendly!
>
> There is a learning curve to all of this stuff. Just take your time
> and bang away at it..
Hi Kevin,
Regards www.geocities.com/dmitrynizh/tubeparams_image.htm: I'm glad
you recommended it to Nick and also offered some practical tips,
thanks. Usability is still suboptimal. At the bottom of the webpage, I
describe some possible improvemnts for the package. It looks as
allowing the jar to be invoked via "java web start" and providing
standard file dialog windows for input/output would greatly help those
lacking java command line experience. I'll try to improve that.
Regards screenprint32: I personally use LVPro to capture the screen
and save the result as a GIF file, but it is good that there is an alternative.
I found the link you posted currently timeouts, though. PC World Downloads still has
the installation at http://www.pcworld.com/downloads/file/fid,22745-order,1-page,1-c,alldownloads/description.html
-Dmitry
www.geocities.com/dmitrynizh
Hi Dmitry,
I think it is an interesting tool, and hope to see you develop it further.
I have mainly used paint_kip and have had some trouble matching the characteristic curves of tubes like the D3A. Is there a possibility (or does one even exist) of using a more complex model that more accurately captures the bending in the origin ends of the plate curves?
Another super nice feature would be a way to tell paintcc just how many characteristic curves you are attempting to capture. (Say you have six curves spaced 0.5V apart for an example.) This would make it a lot clearer when attempting to match the curves.
So far for pentodes you are the only game in town.
A web based interface would be awesome and I was going to suggest that in one of my last posts, but didn't. Starting paintcc and being prompted for a specific gif file to load would also be a cool feature.
A friend of mine recommended your tools to me, and I use both paintcc and a competitor: curve captor. (runs in tcl)
Odd about the link to screenprint32 - worked the last time I tried.
I think it is an interesting tool, and hope to see you develop it further.
I have mainly used paint_kip and have had some trouble matching the characteristic curves of tubes like the D3A. Is there a possibility (or does one even exist) of using a more complex model that more accurately captures the bending in the origin ends of the plate curves?
Another super nice feature would be a way to tell paintcc just how many characteristic curves you are attempting to capture. (Say you have six curves spaced 0.5V apart for an example.) This would make it a lot clearer when attempting to match the curves.
So far for pentodes you are the only game in town.
A web based interface would be awesome and I was going to suggest that in one of my last posts, but didn't. Starting paintcc and being prompted for a specific gif file to load would also be a cool feature.
A friend of mine recommended your tools to me, and I use both paintcc and a competitor: curve captor. (runs in tcl)
Odd about the link to screenprint32 - worked the last time I tried.
Here's a link to the new provtech website where you can download the latest version of screenprint:
http://www.provtech.co.uk/software/screenprint32.asp
It's completely free for home/non commercial use.
http://www.provtech.co.uk/software/screenprint32.asp
It's completely free for home/non commercial use.
Actually it works surprisingly well. The models are only as accurate as the curves you're basing them on of course and with the particular limitations of the model chosen as well.
One other obvious thing is that modern production tubes may differ from the plate curves of vintage versions, and where possible it is better to download the curves for the brand you are actually going to use. Both JJ and Svetlana Winged C provide that information on their websites for most current production types.
One other obvious thing is that modern production tubes may differ from the plate curves of vintage versions, and where possible it is better to download the curves for the brand you are actually going to use. Both JJ and Svetlana Winged C provide that information on their websites for most current production types.
kevinkr writes:
> I have mainly used paint_kip and have had some trouble matching the
> characteristic curves of tubes like the D3A. Is there a possibility
> (or does one even exist) of using a more complex model that more
> accurately captures the bending in the origin ends of the plate
> curves?
>
Kevin,
Paint_KIT uses "Koren Improved" triode model (hence the abrreviation
"KIT"), which for most triodes allows to find fairly accurate
models, given good plots as input. Exceptions are triodes with ultra-low B+ or very substantial or
space-charge effect (example: 6GM8/ECC86) or with distinct "kneebend"
on plate curves due to electrode geometry model, and even these can be
modeled well but it is harder to do. BAck to you question: the
bendings in the origin are mainly due to contact potential and space
charge effects. Try applying positive VCT and combine that with some
KVB variations... If you wish I can try 63A myself later and share the result.
In general, taking care of the origin region is easier than matching
the high_B+/low_current region for some tubes with bad geometry. For
such triodes the model with varying Ex,
http://www.geocities.com/dmitrynizh/paint_pexgt.jar, can be helpful.
Some plots from old tube datasheets are simply bad as an input. I've
briefly looked at the DA3's plot in triode mode on Frank's tube site
and I see that I can model that pretty easily, however this particular
plot is incorrectly drawn. Real plate curves would not show random
changes in density of gridlines that we see there right in the middle.
Hope this helps,
Dmitry
> I have mainly used paint_kip and have had some trouble matching the
> characteristic curves of tubes like the D3A. Is there a possibility
> (or does one even exist) of using a more complex model that more
> accurately captures the bending in the origin ends of the plate
> curves?
>
Kevin,
Paint_KIT uses "Koren Improved" triode model (hence the abrreviation
"KIT"), which for most triodes allows to find fairly accurate
models, given good plots as input. Exceptions are triodes with ultra-low B+ or very substantial or
space-charge effect (example: 6GM8/ECC86) or with distinct "kneebend"
on plate curves due to electrode geometry model, and even these can be
modeled well but it is harder to do. BAck to you question: the
bendings in the origin are mainly due to contact potential and space
charge effects. Try applying positive VCT and combine that with some
KVB variations... If you wish I can try 63A myself later and share the result.
In general, taking care of the origin region is easier than matching
the high_B+/low_current region for some tubes with bad geometry. For
such triodes the model with varying Ex,
http://www.geocities.com/dmitrynizh/paint_pexgt.jar, can be helpful.
Some plots from old tube datasheets are simply bad as an input. I've
briefly looked at the DA3's plot in triode mode on Frank's tube site
and I see that I can model that pretty easily, however this particular
plot is incorrectly drawn. Real plate curves would not show random
changes in density of gridlines that we see there right in the middle.
Hope this helps,
Dmitry
kevinkr writes:
> Another super nice feature would be a way to tell paintcc just how
> many characteristic curves you are attempting to capture. (Say you
> have six curves spaced 0.5V apart for an example.) This would make it
> a lot clearer when attempting to match the curves.
Kevin,
It is easy to add. I thought about this early on but then I got used
to adjusting grid steps - say, when the plot on GIF image has 20
curves and the step is 1V, I specify 2 as grid step to cover the full
grid voltage range without increasing the number of "painted"
gridlines.
> So far for pentodes you are the only game in town.
>
> A web based interface would be awesome and I was going to suggest that
> in one of my last posts, but didn't. Starting paintcc and being
> prompted for a specific gif file to load would also be a cool feature.
I agree. Maybe it is time to find time [pun intended] for hacking
this.
A trick not explained on the webpage in details is as follows: the
tool generates SPICE models that contain a line that looks like this:
MU=77.952 EX=1.5112 KG1=360.0 KP=216.0 KVB=618.0 VCT=0.2 ; Vp_MAX=300.0 Ip_MAX=0.025 Vg_step=0.5
it's sufficient to cut&paste this text into the topmost input field in
the tool's window and press enter, after that the parameters are
parsed and set. To change any of these separately, enter it alone:
KP=200
same can be done for several parameters, of course
VCT=-0.25 ; Vp_MAX=300.0 Ip_MAX=0.02 Vg_step=1
in any order. The semicolon is treated as space but is is a comment in
SPICE files. Additionally, if you type 3 numbers, like these:
300 100 2
these will be interpreted as
Vp_MAX=300.0 Ip_MAX=0.1 Vg_step=2
this is just a handy shortcut.
> So far for pentodes you are the only game in town.
PaintKIP, based on "Koren Improved Pentode" model, while doing good
job, is clearly not as good as needed. There is a need for a model
that allows to add some slope to the curves while allowing to control
the bend. The difficulty is that all that must gradualy merge with the
triode model when the screen grid potential approaches that of the
plate (the bottom slider on PaintKIP allows to see
ultralinear-connection curves: set UL to 0.5:
UL=0.5
and then see it in action.
Dmitry
www.geocities.com/dmitrynizh
> Another super nice feature would be a way to tell paintcc just how
> many characteristic curves you are attempting to capture. (Say you
> have six curves spaced 0.5V apart for an example.) This would make it
> a lot clearer when attempting to match the curves.
Kevin,
It is easy to add. I thought about this early on but then I got used
to adjusting grid steps - say, when the plot on GIF image has 20
curves and the step is 1V, I specify 2 as grid step to cover the full
grid voltage range without increasing the number of "painted"
gridlines.
> So far for pentodes you are the only game in town.
>
> A web based interface would be awesome and I was going to suggest that
> in one of my last posts, but didn't. Starting paintcc and being
> prompted for a specific gif file to load would also be a cool feature.
I agree. Maybe it is time to find time [pun intended] for hacking
this.
A trick not explained on the webpage in details is as follows: the
tool generates SPICE models that contain a line that looks like this:
MU=77.952 EX=1.5112 KG1=360.0 KP=216.0 KVB=618.0 VCT=0.2 ; Vp_MAX=300.0 Ip_MAX=0.025 Vg_step=0.5
it's sufficient to cut&paste this text into the topmost input field in
the tool's window and press enter, after that the parameters are
parsed and set. To change any of these separately, enter it alone:
KP=200
same can be done for several parameters, of course
VCT=-0.25 ; Vp_MAX=300.0 Ip_MAX=0.02 Vg_step=1
in any order. The semicolon is treated as space but is is a comment in
SPICE files. Additionally, if you type 3 numbers, like these:
300 100 2
these will be interpreted as
Vp_MAX=300.0 Ip_MAX=0.1 Vg_step=2
this is just a handy shortcut.
> So far for pentodes you are the only game in town.
PaintKIP, based on "Koren Improved Pentode" model, while doing good
job, is clearly not as good as needed. There is a need for a model
that allows to add some slope to the curves while allowing to control
the bend. The difficulty is that all that must gradualy merge with the
triode model when the screen grid potential approaches that of the
plate (the bottom slider on PaintKIP allows to see
ultralinear-connection curves: set UL to 0.5:
UL=0.5
and then see it in action.
Dmitry
www.geocities.com/dmitrynizh
> Wow your saying triode but I'm using a beam tetrode?
> Problem or not
> Nick
Nick, as Kevin mentioned, you need to use paint_kip.jar file, the link is in section 4 of www.geocities.com/dmitrynizh/tubeparams_image.htm
Everything on this page regards paint_kit.jar applies to paint_kip.jar too, but the model has an additional parameter for the screen grid, thus you can use it for most pentodes and for many tetrodes (pentodes have "better", smoother, more "ideal" plate curves, and that is why the model serves pentodes better than tetrodes). The paint_kip tool also has a slider called "UL" that I mentioned in my previous post, allowing you to see ultralinear loadlines. When UL=1 your pentode model becomes triode.
Dmitry
> Problem or not
> Nick
Nick, as Kevin mentioned, you need to use paint_kip.jar file, the link is in section 4 of www.geocities.com/dmitrynizh/tubeparams_image.htm
Everything on this page regards paint_kit.jar applies to paint_kip.jar too, but the model has an additional parameter for the screen grid, thus you can use it for most pentodes and for many tetrodes (pentodes have "better", smoother, more "ideal" plate curves, and that is why the model serves pentodes better than tetrodes). The paint_kip tool also has a slider called "UL" that I mentioned in my previous post, allowing you to see ultralinear loadlines. When UL=1 your pentode model becomes triode.
Dmitry
Ok I wasn't sure & I was wondering why every one here would call a 813 a pentode when it's not.
I will scan the curves this weekend and put them into paint, and I promise to let you guy's know.
Once I get a model I will have a schem. whipped up quick I'll also throw my hand drawing up here to for the heck of it.
Nick
I will scan the curves this weekend and put them into paint, and I promise to let you guy's know.
Once I get a model I will have a schem. whipped up quick I'll also throw my hand drawing up here to for the heck of it.
Nick
Not sure yet i'm going to do some seriuos work tonight I'll get that model sorted out.
It's kind of dificult during the week because my girlfriends lids are alway's on the computer with the scanner.
But this weekend they are over at there dads so I will have it all to my self.
Do you know of any surpluss electronoic store in Quincy or Boston?
Nick
It's kind of dificult during the week because my girlfriends lids are alway's on the computer with the scanner.
But this weekend they are over at there dads so I will have it all to my self.
Do you know of any surpluss electronoic store in Quincy or Boston?
Nick
nhuwar said:
Well one of the interesting things about the beam power tetrode aka kinkless tetrode was that it was intended to circumvent the (IIRC Philips?) pentode patents. If you look at the characteristic curves for a beam power tube or a pure power pentode in most cases the differences are quite trivial. For the purposes of spice beam power and KT type tetrodes are treated as pentodes. Classical tetrodes with their big kink imo aren't accurately modeled by any available spice model.
IMO The existing pentode models aren't always that good either, and the ones I've made so far are very poor..
I'm still going to get a hand draw schematic up here first I just have to make it fit on a sheet of paper. Iam almost finished with it though it wont have B+ supply in it.
Right now I'm working on the intermeadeate stage but I dont think it's called that in audio so i guess the pa tube drivers.
Still not sure if I am going to go tube or solid state??
Nick
Right now I'm working on the intermeadeate stage but I dont think it's called that in audio so i guess the pa tube drivers.
Still not sure if I am going to go tube or solid state??
Nick
There used to be a place called Eli Hefron somewhere in Cambridge, unfortunately I have not been there in years and I think that if they even still exist that it would mostly be industrial and computer hardware.
I don't know if Abbott Electronics in Woburn is still around. You might want to check.
For the few things I need to get locally I usually go to You-do-it electronics in Needham. Expensive, but they do have a lot of parts.
I purchase most of my parts online.. For audiophile/tube exotica I go to these places:
http://www.percyaudio.com/
http://diycable.com/
http://www.tubesandmore.com/
Ebay!
For everything else either Digikey, Mouser or Newark Electronics.
I am still a sovtek dealer and I tend to buy most of my current production stuff from New Sensor in NYC. (Keep my few remaining clients happy.. )
No one local is really tube friendly, and electronic hobbyists are as rare as buffalo nickels these days so no one caters to them locally any longer.
During the season the MIT Flea is a great source of all sorts of parts, and sometimes tubes as well. (Variable)
Here is the latest list of Fleas in the area (aka formerly known as a "hamfest")
http://web.mit.edu/w1gsl/Public/ne-fleas
I usually hit the one in Westford (Feb 18th) because I love antique tubes - my system uses a few of them.. Focus is on early radios and the tubes and parts that went into them. Some tube hifi as well. I go just for tubes and vintage shortwave radio parts..
I don't know if Abbott Electronics in Woburn is still around. You might want to check.
For the few things I need to get locally I usually go to You-do-it electronics in Needham. Expensive, but they do have a lot of parts.
I purchase most of my parts online.. For audiophile/tube exotica I go to these places:
http://www.percyaudio.com/
http://diycable.com/
http://www.tubesandmore.com/
Ebay!
For everything else either Digikey, Mouser or Newark Electronics.
I am still a sovtek dealer and I tend to buy most of my current production stuff from New Sensor in NYC. (Keep my few remaining clients happy.. )
No one local is really tube friendly, and electronic hobbyists are as rare as buffalo nickels these days so no one caters to them locally any longer.
During the season the MIT Flea is a great source of all sorts of parts, and sometimes tubes as well. (Variable)
Here is the latest list of Fleas in the area (aka formerly known as a "hamfest")
http://web.mit.edu/w1gsl/Public/ne-fleas
I usually hit the one in Westford (Feb 18th) because I love antique tubes - my system uses a few of them.. Focus is on early radios and the tubes and parts that went into them. Some tube hifi as well. I go just for tubes and vintage shortwave radio parts..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.