Nope. SW is not backwards compatible across major versions (it is
between service packs within a major release).
This is a major item of discussion that people want.
You can save back from later versions of SW to earlier ones by saving
as a Parasolid. You will get a "dumb" solid when you open up in the
old version, without history or parametrics.
The best you can do is save a part in Parasolid and open in a previous
version of SW and use FeatureWorks to try and recognize some geometry
and pull out some history. This only works on simple geometry though.
This makes playing in the same sandbox with suppliers/customers with
different versions a little tricky.
I know it's not exactly the same thing, but my Word XP will save as other
versions back to '95 or even DOS. It will occassionally warn you that some
features may be lost, but it at least it tries.
It is interesting that Solidworks purports to do the same thing for Autocad
files. I agree that the SW files may be more complex, but that doesn't sound
impossible. Imagine what a company could do working with it's own files, not
The biggest difference here (and it's huge) is that if your missing
something in a word doc, or even an Autocad 2D drawing, the system can still
usually read, and use, what's there.
Solid modeling systems in general are much different. The mathematics of the
model "must" resolve to form valid topology within the accuracy of the
system. With SW that's at least 10e-8 meters, and this is just the topology,
faces, intersections, and vertices.
Add to that the construction database with it's dimensions, geometric
relations, equations, and all the rest, and you have an exremely complex
If the system can't resolve it, (missing, corrupt, incomplete, or wrongly
defined elements) alot of things can happen. All of them bad by the way. At
best you could get a "corrupt file" warning, and it just won't open. At
worse, it could freeze or crash the application. Or even crash the entire
system to a blue screen.
These are just some of the requirements for a model within the software
version in which it was created. Now you want to save this binary hairball
to an earlier binary hairball that doesn't support some of the critical
elements that comprise it's construction in the newer version.
I don't doubt that if SW were to throw enough money and development time at
it, they could make something that worked in "some" cases. The problem is,
they would take so much heat for the times it didn't work, that they'd just
prefer not to open this can of worms in the first place. Can't say as I
blame them either
You make some interesting and valid observations, but I would suggest that a
partial solution to this problem would serve SolidWorks much better than
their current approach. Since each new release of SolidWorks adds new
features, constructs, etc. it would be appropriate if attempting to save any
of the newer items in an old version would produce an error/warning that the
new entities are not supported. In these cases, SolidWorks could gracefully
notify the user of the problem and not produce the desired old version file.
On the other hand, for geometry that doesn't require the newest features,
entities, etc. SolidWorks could produce the desired old version easily. I
would venture to guess that at least half of all parts created would fall in
this later case and the ability to save them in old versions would be very
beneficial to SolidWorks users. Since cases involving the newer
features/entities would be somewhat frequent as well, these would serve as
an impetus for all users to stay current with SolidWorks. The biggest
advantage of such an approach would be the change in user perception of
SolidWorks Corp. Some of the people that believe the current approach (or
non-approach) to this problem is a manipulative way to force upgrades would
reconsider if SolidWorks made an attempt to deal with older SolidWorks
If there were no valid reasons for a company to run an older version of
SolidWorks, this topic would be strictly academic. Unfortunately, there are
valid reasons to run older versions. Stability is one reason often
discussed on this newsgroup. I won't say anything more about this reason,
except to note it is one reason. Additionally, there are costs associated
with changing versions other than the basic fees paid to SolidWorks. For a
large company, changing versions requires much coordination, training,
upgrade labor, etc. Such companies don't necessarily choose to go through
the whole process once a year. They choose to minimize the time spent on
upgrades in order to improve their efficiency and minimize their overall
cost, despite paying all of their maintenance fees to SolidWorks. I am
familiar with several large companies (not necessarily using SolidWorks)
that limit the introduction of new versions for this very reason. Often,
they skip some releases because the new releases come out so frequently.
I believe that jss87 is correct when he wrote "Imagine what a company could
do working with it's own files, not someone elses'". SolidWorks should be
able to save their their basic geometric constructs in slightly older
version files with only minimal effort. I would venture to guess that each
new release of SolidWorks does not involve the complete re-definition of the
file formats. For the record, I would only expect SolidWorks to support one
or two years of older SolidWorks versions.
I agree that it should be very easy to do some prismatic parts. Many would
only require transposing the instructions. In fact, all you would need are
the instructions in some cases. SW 95 didn't store any hard geometry in the
part file. The files were really small too. Took longer to load though.
The place you would run into problems with prismatic parts are things like
feature patterns. For instance SW2004 can't do true 3D patterns along a 3D
curve. 2005 & 2006 can. How do you account for this ??
"Making unsupported features dumb solids" kind of rolls off the tounge real
easy. Not so easy when you really think about it. If the base features of a
shape are two or three complex lofts or sweeps, shelled with a ton of bosses
ribs and other internal stuff, most of it will be orphaned if the lofts are
dumb. Whether they're left dangling or broken, allot of work will be
necessary on the recieving end to fix it up. How many time can a model make
the round trip between the newer/older-older/newer cycle before the only
real features left are a couple of holes ??
I really don't think it can be implemented in a way that has any real value
I assume your final statement of " I really don't think it can be
implemented in a way that has any real value." is because of the current
ability to export SolidWorks files as dumb solids via parasolid, IGES, etc.
That seems to be a reasonable assessment, particularly when you consider all
of the other issues that can be addressed by SolidWorks.
I'm curious, do you completely discount the theory held by some CAD users
that backward compatibility is not supported because companies like
SolidWorks wish to "encourage" users to always use the newest version?
Personally, I have heard many people express their concerns about this issue
and I have wondered about the validity of their arguments. To be honest, I
don't have a strong opinion either way on the topic due to my uncertainties
about the big picture.
I haven't post anything in this group for a while.... But I love when
this subject comes up. It's so much fun reading
everyone's ignorant responses. The irony here is, I know I'm sounding
ignorant myself but I can't hold back from this
subject without sounding impolite or sarcastic....
Yes guys, it's true. Solidworks and other companies have created a
giant money generating conspiracy. Together,
they make you believe they can just quickly generate some simple code
that allows you to save to a previous version.
As a matter of fact, it would be so easy to do they would rather just
make you pay for upgrading to the newer software.
(Am I sounding sarcastic here... I hope so!)
Apparently nobody understands natural order, programming, or even logic
for that matter. What makes people think Solidworks
can just have 1 guy hand over the code to make backwards compatibility
happen? The truth is that it would be way too complex and big of a
problem from a programming/database standpoint to be able just to be
readily available at the same cost the user. I have to believe this
would cost a fortune in development only to meet it's demise when
people won't pay $$$ for the ability to use it. THIS DOES NOT COME FOR
When you think about all the problems and money it would take to
accomplish such a task.... doesn't it just make sense to pay for the
upgrade? Apparently some people just don't understand this.
By the way.... I keep putting my DVD movie into my CD player I bought
12 years ago and it doesn't work.
Why can't they just make the DVD backwards compatible???? GIVE ME A
Sorry for sounding rude... I meant to sound realistic.
Don Van Zile
If someone chose to do it, it is certainly doable. Barenboym had
developed something that worked on limited feature types, but it worked.
Look at what SW can do with native ProE files. If it can be done from
ProE to SW it can certainly be done much more easily from SW2005 to
SW2001+. Any features that don't have an equivalent in an old version
are either brought forward as a dumb solid or left out, same as the ProE
conversion. Something is better than noting.
The question then becomes "why". It probably wouldn't be a big money
maker, because the people who are hanging out on old versions are
probably doing it to avoid spending money in the first place.
I honestly don't believe it is a technical issue. It's a business
You've correctly pointed out that there would be very real limits.
Solidworks doesn't really do much with Pro files based on complex shapes.
Your choices are single body, or a mess. The mess isn't worth trying to
clean up. So much for dumb or incomplete features. It does work on prismatic
parts "provided" the person doing the modeling (in pro) used straight
forward construction methods. I've seen Pro models where the guy used
variable section sweeps to construct countersinks.
Something is better than nothing is only "your" perspective. There will be
legions of users who will piss and moan because of the limitations. They'll
complain loudly about another half baked Solidworks feature. It's human
nature, not worth the hassle.
"Your choices are single body, . . ."
Here's my take on this:
1) It's parasolids in there somewhere.
2) If we can take in pro files as a dumb solid, at least give us the
option to take in a "frozen" parasolid from future SolidWorks versions.
Predicated on compatible parasolid versions maybe (not something I can
3) Given a drawing, at least let the editor open a future version in
view mode, or perhaps even a variant of "detached drawings". I know
the viewer is available, but that's not my point.
Full backwards compatibility would be a total CF.
Today I had a model from my vendor in 2005 and could not open it in
2004. OK that makes sense, BUT it would be nice if I could open the
model as a "frozen" solid, basically having the editor read the
internal parasolid data for me - basically nullifying the need for a
dumb solid as the communication vehicle between solidworks versons. I
think that this could even extend into an assembly being able to take
in future data as frozen.
I think that being able to open a file as frozen and use the geometry
as needed is not too much to ask for. Full parametric compatibility is
probably a very tall order and strewn with roadblocks.
I hate to even mention this (surely opening the floodgates of idiocy)
but what about having a limited direct manipulation mode when opening a
future file in an older version?
And then you have NX which can take a totally dumb file and treat
features in it as if they were parametric in origin, dragging,
retrimming, etc. 90% of users use a very simple pallette of features.
Those features haven't changed much over time. And from what I
understand the code to make them in the older versions remains in the
newer versions to enable backward compatibility.
That's what I was thinking. Something is wrong and maybe this is too
big a leap. To have history trees and direct co-operate.
But damn what a wish list item.
I expect them (or someone) to come up with a solution to make this sort
of big picture stuff happen. I hope that they have a think tank and
would not be shocked if there is some sort of body in the company that
thinks about this stuff all day. Sort of like an advanced development
function. I know that we will likely wait a decade for a major change
and perhaps SolidWorks will not be capable of being the "lusted after"
CAD of the future. They may just rest on their laurels and get comfy
maintaining the status quo.
In any case, the future belongs to the people who can innovate. In CAD
and any other business.
Your observations about the "something is better than nothing" approach make
alot of sense. Considering the number of complaints I hear from SolidWorks
users, any new functionality that is less than complete will probably lead
to as much criticism as praise. Maybe SolidWorks would be better off
leaving this topic unaddressed.
I think this reaction is kind of severe. It's true that no one likes
half baked features, I'm the first to jump on that wagon too, but the
translation issue has never been a 100% success. IGES has been half
baked for a long, long time, but people still use it even though there
are better alternatives. The ProE translator is far from perfect too,
but if I have to rebuild a ProE part, I'd rather start with maybe 80% of
the sketches than nothing.
I suppose if you're so down on things that don't work perfectly, you've
never used FeatureWorks. FeatureWorks can strip out the features it can
recognize and leave the rest as a solid imported blob with parametric
features around it. I'm not suggesting that anyone should use FW to
translate new SW files, but I don't see why it wouldn't be technically
feasible for a program that looks at the instructions to make the
features to recreate them rather than looking at the geometry itself,
and use the parasolid data for features that don't exist in the older
version. You'd wind up with a blob that represents the shape feature,
and paremetrics for other things that existed in that version.
Link is still active after a couple years. The help file is in Russian,
which, maybe I'm wrong about this, but isn't really helpful. Anyway,
Barenboym started the project and seems to have dropped it for some
reason. I know we traded a couple of emails, and the intention was to
go forward with more feature types. This version I think did extrudes,
revolves and simple fillets.
Technically difficult? Probably. Impossible? No.
Financially feasible? Probably not.
Financially for SW it would be a big loser because obviously fewer
people upgrade if they're not forced. I don't believe SW will ever
provide this. A 3rd party provider like Barenboym would be the folks
who would have to do it.
There's no way of knowing, but someone might have exerted some pressure
to stop that project too... You can get to Framingham from Concord in
about 30 minutes.
I don't think it will ever become a reality, but I think it's definitely
Despite the postings from a few individuals that "know" the truth about the
technical difficulties associated with this issue (their postings are
unsubstantiated, coincidentally), Matt makes a very important point.
If anyone can offer more than just claims to support the concept that its
technically not feasible to support older versions, please provide your
information. I am certain many people would be interested in learning about
Its interesting to observe the accusations of ignorance directed at users
asking relevant questions and discussing those issues. Is such behavior the
result of arrogance, impatience with other users, or some other cause?
Oooooooooooooo,,,,, it's a conspiracy... gimme a break.
I haven't called anyone ignorant, all I've asked you to do is really think
about it. Another example. SW has changed it's surfacing algorithms in every
releas since 2001+. I have several product models that were done in 2001+
that I can't even bring "FORWARD". Why ? because the surfaces change shape,
some have failed features asa result, and "ALL" of them no longer match the
very expensive production injection molds.
Can you imagine trying to go backwards to a version that didn't support
things like local curvature and tangency control. The resulting mess would
I have a feeling that maybe you just do prismatic mechanical stuff. Nothing
wrong with that, but it doesn't give you a very good view of the larger
Coincidentally, the whole purpose of this thread was to get people thinking
about it and commenting accordingly. I'm sure we all agree that's the
I have seen ample examples of SolidWorks files that can't be brought forward
without user intervention. Sometimes loading older files results in
distinct errors and in other cases there are slight differences in the
geometry. Most people that I know view this as a limitation/shortcoming of
SolidWorks, but we deal with it and some of these situations are unavoidable
(i.e. not a reason to criticize SolidWorks).
Yes, I can imagine going backward to a version that didn't support things
like local curvature and tangency control. This problem is dealt with
routinely when we export files as dumb solids in any number of older IGES,
parasolid, STEP, or ACIS versions. Reading through the various postings in
this thread, a variety of people have imagined it and offered their thoughts
about how SolidWorks might deal with such situations.
That would be a bad assumption, but that's another story...
Of course, I do work with alot of prismatic shapes (as many people do) and
the ability to save them as older versions of SolidWorks files would be
useful for a variety of reasons.
Thanks for your interesting postings on this topic.
I just ran a test using the ExchangeWorks v.1.4 conversion utility that
Baren-Boym was developing. I re-confirmed my earlier findings that even the
simplest geometry doesn't convert accurately.
I modeled a simple 1.0" diameter cylinder and saved it as a 3XF format file
using ExchangeWorks. I then imported the 3XF file back into SolidWorks
using ExchangeWorks. A quick examination of the dimensions showed that the
diameter had been changed to 0.99999995". Not a significant difference, but
certainly nothing to ignore.
Autodesk attempted to address this type of issue with their "proxy object"
approach. It would be kind to say it doesn't work very well.
While backwards translation would help me in some cases, I want it to be
robust, and allow for round-tripping without data loss. What many users
have suggested in this post would alter the file to such an extent that
design intent and future editing would be negatively compromised.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.