X-Authentication-Warning: delorie.com: mail set sender to geda-help-bounces using -f X-Recipient: geda-help AT delorie DOT com Message-ID: <5468F15C.2030907@hallowell.com> Date: Sun, 16 Nov 2014 13:47:56 -0500 From: Chris Arena User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.2.0 MIME-Version: 1.0 To: geda-help AT delorie DOT com Subject: Re: [geda-help] Multiple supply voltages References: In-Reply-To: Content-Type: multipart/alternative; boundary="------------070601030304090709020601" X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - server.sohouniverse.net X-AntiAbuse: Original Domain - delorie.com X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12] X-AntiAbuse: Sender Address Domain - hallowell.com X-Get-Message-Sender-Via: server.sohouniverse.net: authenticated_id: chris AT hallowell DOT com X-Source: X-Source-Args: X-Source-Dir: Reply-To: geda-help AT delorie DOT com This is a multi-part message in MIME format. --------------070601030304090709020601 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Mike, An interesting factoid is that many fields --- don't know which ones --- are parsed as expressions, so names with dashes, or pluses or other arithmetic operators might not work as you might implicitly expect. I don't know if the parse is for arithmetic or regular expressions. Lesson: avoid something that resembles an arithmetic operator in any field for a name; underscores are a good substitute. It would be great to have an explanation about fields that are parsed for expression because it wasn't intuitive to me at all that was going on when I ran aground on that problem. I admit I'm not sophisticated enough to see how that sort of feature is useful. Examples of its intended use would also shed some light on its utility. Does anyone have some? Chris On 11/16/2014 01:26 PM, Mike Bushroe wrote: > Sorry, forgot to post my current config with the last post. > > gschem 1.8.2 (g875406c) > gsch2pcb 1.6 > xgsch2pcb 0.1.3 > pcb version 20140316 Compiled on Aug 29 2014 at 07:56:58 > Ubuntu 14.10 (just up from 10.04 through 12.04) > Toshiba 5.7GiB Core 2 Duo 2.2GHzx2 64bit > > I got an impossible to get around error when using the latest gschem, > sch2pcb, and pcb that would load onto 10.04, and then it would delete > the .new.pcb file with the error before I could read it and try to > figure out what the error was and how to fix. Never did, so I was > forced to finally do the system upgrade. I still hate Unity. But at > least now the hidden error is gone, leaving only a few thousand > obscure ones left! > > I am trying to make a single phase power to 3 phase power converter > for a friend to run a lathe, and throwing in partial speed control and > reverse for free. As a result, I am using two full wave diode blocks > to generate plus 120V DC and minus 120V DC for the MOSFET H-bridges to > chop up back into 220V AC. But that means the nmos switches have to > sit on the lowest voltage, the -120V supply rail, and the driver chip > needs 15V above the -120V, and the digital inputs to the drivers need > 5V above the -120V rail. That means a 7805 and a 7815 whose ground > pins are not at chassis ground. On the other hand the ATMega that I am > using to control the whole mess does not have the main digital power > and ground pins explicitly on the symbol and so aren't hardwired into > the second 7805 that provides true 5V above ground. > > I keep getting warning messages about nets being shorted together, > which I think means that hidden attributes in the 7805 and 7815 symbol > are shorting -120V to chassis ground. On the other hand the buried > power and ground connections on the Atmel seem to be making Vcc pin > not found errors. > > I have tried copying the 7805 and 7815 symbols into my circuit > directory and making local copies of them and removing all references > to 'gnd'. But then I get found multiple instances errors, so I am > guessing that it does not do 'local instance over rides global > instance' type symbol loading and I add a dash number to the local > name and re-edit the schematic to use the new name. I tried many ways > and many power symbols to get a Vcc net going that the hidden pins on > the ATMega would use, but that seemed to cause more problems than it > cured. > > My current guess is that I should go back to the original delivered > symbols and find some way to override the built in hidden pin net > bindings and either have ignore everything and just meekly go with > where the wiring connects it or find a way to uniquely name my unusual > power rails (see post above about trouble with names) and get them to > connect to the 'ground of my choice' instead of the chassis ground. > But I don't recall seeing any tutorials on how to over ride the hidden > 'net:' pin attributes other than that I should 'promote' the hidden > attribute and change it to something. Can anyone point me to a > different tutorial that gives more specifics on overriding power > connections? (when 'ground' really means '-120V'?) > > Thanks again fro your help! > > Mike > -- > "Creativity is intelligence having fun." — Albert Einstein -- Chris Arena Hallowell EMC Pittsfield, MA 01201 413 445 4263 (work) 401 293 0449 (home) 757 831 9255 (cell) --------------070601030304090709020601 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 8bit Mike,

An interesting factoid is that many fields --- don't know which ones --- are parsed as expressions, so names with dashes, or pluses or other arithmetic operators might not work as you might implicitly expect. I don't know if the parse is for arithmetic or regular expressions.

Lesson: avoid something that resembles an arithmetic operator in any field for a name; underscores are a good substitute.

It would be great to have an explanation about fields that are parsed for expression because it wasn't intuitive to me at all that was going on when I ran aground on that problem. I admit I'm not sophisticated enough to see how that sort of feature is useful. Examples of its intended use would also shed some light on its utility. Does anyone have some?

Chris

On 11/16/2014 01:26 PM, Mike Bushroe wrote:
Sorry, forgot to post my current config with the last post.

gschem 1.8.2 (g875406c)
gsch2pcb 1.6
xgsch2pcb 0.1.3
pcb version 20140316 Compiled on Aug 29 2014 at 07:56:58
Ubuntu 14.10 (just up from 10.04 through 12.04)
Toshiba 5.7GiB Core 2 Duo 2.2GHzx2 64bit

I got an impossible to get around error when using the latest gschem, sch2pcb, and pcb that would load onto 10.04, and then it would delete the .new.pcb file with the error before I could read it and try to figure out what the error was and how to fix. Never did, so I was forced to finally do the system upgrade. I still hate Unity. But at least now the hidden error is gone, leaving only a few thousand obscure ones left!

  I am trying to make a single phase power to 3 phase power converter for a friend to run a lathe, and throwing in partial speed control and reverse for free. As a result, I am using two full wave diode blocks to generate plus 120V DC and minus 120V DC for the MOSFET H-bridges to chop up back into 220V AC. But that means the nmos switches have to sit on the lowest voltage, the -120V supply rail, and the driver chip needs 15V above the -120V, and the digital inputs to the drivers need 5V above the -120V rail. That means a 7805 and a 7815 whose ground pins are not at chassis ground. On the other hand the ATMega that I am using to control the whole mess does not have the main digital power and ground pins explicitly on the symbol and so aren't hardwired  into the second 7805 that provides true 5V above ground.

  I keep getting warning messages about nets being shorted together, which I think means that hidden attributes in the 7805 and 7815 symbol are shorting -120V to chassis ground. On the other hand the buried power and ground connections on the Atmel seem to be making Vcc pin not found errors.

  I have tried copying the 7805 and 7815 symbols into my circuit directory and making local copies of them and removing all references to 'gnd'. But then I get found multiple instances errors, so I am guessing that it does not do 'local instance over rides global instance' type symbol loading and I add a dash number to the local name and re-edit the schematic to use the new name. I tried many ways and many power symbols to get a Vcc net going that the hidden pins on the ATMega would use, but that seemed to cause more problems than it cured.

  My current guess is that I should go back to the original delivered symbols and find some way to override the built in hidden pin net bindings and either have ignore everything and just meekly go with where the wiring connects it or find a way to uniquely name my unusual power rails (see post above about trouble with names) and get them to connect to the 'ground of my choice' instead of the chassis ground. But I don't recall seeing any tutorials on how to over ride the hidden 'net:' pin attributes other than that I should 'promote' the hidden attribute and change it to something. Can anyone point me to a different tutorial that gives more specifics on overriding power connections? (when 'ground' really means '-120V'?)

Thanks again fro your help!

Mike
--
"Creativity is intelligence having fun." — Albert Einstein

-- 
Chris Arena
Hallowell EMC
Pittsfield, MA 01201
413 445 4263 (work)
401 293 0449 (home)
757 831 9255 (cell)
--------------070601030304090709020601--