Time to move away from Altium?

[tesla500]

I've recently started running up against the limits of Altium for a project I'm working on. I'm routing a DDR3 interface, and the high speed design rules are quite inadequate. Specifically, there's no support for matched net lengths for the flyby topology used by the DDR3 address bus, as well as no support for compensating for die-to-ball trace lengths on IC packages. I've had to resort to routing the DDR3 interface in sections (processor to memory, then memory to memory), and matching each individually. This works, but renders the DRC useless once the non-matched traces are routed to the termination resistors.

In addition to the lack of some critical features, others have raised concerns about the slowness of bug fixes, and Altium's focus on useless features like their FPGA tools and this new Vault/cloud thing.

The main alternative I've found that has the features I need is Cadence Allegro. This package seems to be used for pretty much every FPGA or SOC dev board I've seen. The main difficulty in switching is porting over my Altium library and recreating the schematics, as I doubt there are any importers for Altium files.

Any thoughts on this?

[yanir]

I too find myself disappointed with altium frequently. Being new to fpga design I bought into the whole innovation station concept (but I bought a nb3000, and not the ridiculously expensive nanoboard.) unfortunately the examples designs rarely work and their software stacks are full of bugs. I didn't renew my subscription because of that.

Cadence allegro is a way more expensive package then altium and less intuitive to learn. I have a seat of the entry level package at work and prefer altium for my designs. I however don't route ddr3 tracks, so my needs are different.

I imagine that the allegro high speed module is a pricey one but you'll have to do the math for your self and see it pays to get it.

I'd hate to redo all of those footprints tho. If altium actually succeeds in the whole unified footprint vault idea it would be great.

 Now if only it wouldn't crash every time I try to save...

[joelby]

I have no idea if there are already any solutions to solve your problems specifically, but maybe you could add on support for your high speed routing needs by writing scripts?

There are a whole bunch of third-party scripts hosted here - at the very least, they might give you some inspiration and code examples if you feel like doing some programming.

[EEVblog]

The DDR length stuff has been considered for many years now, and has been looked into as they have publicly acknowledged numerous times. AFAIK, you won't see it anytime soon, but I could be wrong, a lot of people have been asking for real signal integrity stuff for years, and as everyone knows, they have released practically squat in that area.
I don't know what's in the pipeline, but I doubt a real SI solution will come any time soon.

IMO they now have a massive resource problem, having dumped a large part of the staff in the delusional China strategy.
That's on top of, IMO, not enough focused resources before all the China thing happened when staff were at their peak.
So I can't possibly see how they can bring out all the new core stuff people want, real signal integrity stuff is not easy to do.
You would have noticed most bugs fixes and enhancement recently have been core PCB/SCH related which is good, they seem to be trying to keep the faithful happy on that front.

Yes, the whole FPGA concept they spent a decade and $100M+ building up has fallen into a heap as most people predicted. They were biting off far more than they could chew there. Lots of nice concepts in it, and works very well in some areas, but "some" isn't good enough.
They will never get there with their FPGA solution, of that I am certain. You only have to look at the lack of new device support over the years. Many promises, and not much delivered.

The coders they have left are very good, but you can only do so much with so few.
Maybe their China strategy has worked, and they were able to hire a ton of talented programmers very cheaply and get them up to speed, and dedicate them to the core PCB/SCH areas, and we'll see some fantastic new core tools out of the pipeline soon. I don't know what's happened since I've left, but I think my flying pig is more likely.

It's a real shame, because the tool is awesome, and could be the #1 PCB/SCH tool in the world, if only they would drop all the blue-sky all things to all people dreaming. But alas, they have shot themselves in the foot at practically every turn.

Dave.

[bfritz]

With the complexity of the projects you are doing, you will appreciate Cadence products.  There is no other choice for the feature set they offer.  I've worked at several different IC manufacturers, and two different big name computer manufacturers, and all have used a variety of Cadence products.

Honestly, if Altium were half the price, they would be great for 90% of users.  With Altium's price point where it is, I think anyone interested in better off with a Cadence solution.

[EEVblog]

Honestly, if Altium were half the price, they would be great for 90% of users.  With Altium's price point where it is, I think anyone interested in better off with a Cadence solution.

At least it's much better than what they used to charge at their peak, around $10K-12K, with the Nanoboard NB2 at $4K.
They wised up and did the huge price drop, and then have slowly creeped back since then, but still a long way from 5 digits.

Dave.

[Blue]

Hi Guys,

Looking at the share price I can understand the management a bit. They had to do something back in 2008/2009 since their share price dropped dramatically.
But the move they made is a rather stupid one. They should have focused on what they were good at, just what Dave said. And they should have shelved the whole FPGA and vault stuff. But their management was/is shortsighted for the company. For them personally perhaps it was a wise move: Increased share price over the last months was great. Now they can dump their shares/options and retire. Gosh, I hate that mentality...

Cheers,

[EEVblog]

Looking at the share price I can understand the management a bit. They had to do something back in 2008/2009 since their share price dropped dramatically.
But the move they made is a rather stupid one. They should have focused on what they were good at, just what Dave said. And they should have shelved the whole FPGA and vault stuff. But their management was/is shortsighted for the company.

Managements problem is not actually shortsightedness, it's being too far-sighted.
They thought FPGA were the future and they could pull it off. They failed, terribly.
They thought $4K nanoboards there the future and they could pull it off. They failed, terribly.
They now think Vault technology and the "internet of things" is the future and that they can pull it off. They will fail miserably again, practically guaranteed.

The only thing that has kept them going for more than a decade is their core PCB/SCH tool, which is all almost all of what the customers care about and use. Yet management have had seemingly zero interest in being the world's #1 PCB/SCH tool, they had to focus on the FPGA/embedded world, and now the cloud and "internet of things". The results was obvious to everyone, the core PCB/SCH tool had very little development or focus behind it, so it now has no decent SI solution for example.
And all the customers have had to pay for all the stuff they don't want or need hoping they'll pick up a few crumbs on the core PCB/SCH side.

The funny thing is, they are not technically wrong in any of their visions, it's just that Altium as a company can't make the slightest dent in those areas, it's just not what the company does or is good at or capable of, or what their customers want from them.

For them personally perhaps it was a wise move: Increased share price over the last months was great. Now they can dump their shares/options and retire.

Hardly.
Considering that the share price was upwards of $5 at one point (with less shares in the pool), even doubling the current share price is pretty pathetic.
The biggest share holder is the original founder, and based on say a share price of 20cents, he'd cash out with only 7 digits, less money than what Farnell paid for Eagle. Not much after 25 years of hard work and a company that was once worth 1/4 to a 1/2 billion dollars.

Dave.

[Greg J]

The only thing chinese will do, is make a pathetic copy of the software and run around with it. Not a smart move really, the whole moving development to far east/asia .
Might work factory wise, but software wise - from experience - it always goes tango uniform.

I hope you don't count much on your share options They should make that sort of thing illegal..

[JuKu]

On the other hand: For FPGA, the 300€ Nanoboard is probably the best value for money there is on that field. Also, it is crucial to have the schematics/PCB closely tied to FPGA side, for pin swap, routing constraints etc, and it is nice to be able to use the same interface (schematics, block design etc) for all design entry.  Altium does fine here. But IMO, this is where it should stop, design entry. It is foolish to think they can keep up with all FPGA vendors.

It is sad to see that they are loosing ground on their core competence: Altium is no longer the adequate for modern designs. They are severely handicapped (compared to others) is high speed routing, design validation (or example, Altium lets you connect 3v input to 5v output and two FPGA/processor pin outputs together), version management, product design (it is one schematic - one PCB only; cmon, only the simplest products have only one PCB) and so on.

I don't think I'm going to pay the next maintenance next time, I'll start saving for a modern tool.

[EEVblog]

On the other hand: For FPGA, the 300€ Nanoboard is probably the best value for money there is on that field. Also, it is crucial to have the schematics/PCB closely tied to FPGA side, for pin swap, routing constraints etc, and it is nice to be able to use the same interface (schematics, block design etc) for all design entry.  Altium does fine here. But IMO, this is where it should stop, design entry. It is foolish to think they can keep up with all FPGA vendors.

Yes, we designed the NB3000 to be very cost competitive.
The included bundle with the 12 month license is actually an exceptional deal for FPGA development. IF it suits your purposes and the Altium package can ultimately do what you want, then it's a great way to quickly develop a demonstrator project.
The Altium FPGA solution is one of the nicest if not the easiest to use FPGA solutions on the market.
The big BUT of course is it effectively locks you into the Altium system, once that 12 month license runs out, you have to fork over the $$$$ or you are toast. Then you have to hope the Altium environment can do exactly what you want, otherwise you are forced back to the vendor tools and lose the advantage of the Altium development system.

Some customers have made good use of it, but many more have found it wanting when they start to do serious projects or use peripherals not available on the Nanoboards and have been forced to abandon it and start from scratch with other tools.

It is sad to see that they are loosing ground on their core competence: Altium is no longer the adequate for modern designs. They are severely handicapped (compared to others) is high speed routing, design validation (or example, Altium lets you connect 3v input to 5v output and two FPGA/processor pin outputs together), version management, product design (it is one schematic - one PCB only; cmon, only the simplest products have only one PCB) and so on.

I don't think I'm going to pay the next maintenance next time, I'll start saving for a modern tool.

What would you switch to?
The thing that keeps Altium going is that there probably isn't another competing solution in the same price bracket.
Plus the fact that you can get around all the limitations the tool has for high end designs. Often it's easier and quicker to just do the manual work-arounds than buy and learn a more powerful tool from scratch.

I reckon they can hold out for another year or so the way they are going, but if they haven't made major improvement in the core tool by then, and/or tried to capture the low end market back, the users won't be kind...

Dave.

[rz2k]

Might work factory wise, but software wise - from experience - it always goes tango uniform.

IMHO, AD is now dead because it is written in Delphi (!!), you cant find good Delphi guys without ages of HR, it is not just a regular type of programming language, it is abandoned by original designers, sold 3 or 4 times envorinoment with tons of quirks and glitches, in China they will not found them too, by good Delphi guys I mean really good programmers who will understand how to do stuff needed for EDA best way, not just regular "I do specification and nothing more" guys.

[mikeselectricstuff]

The big BUT of course is it effectively locks you into the Altium system, once that 12 month license runs out, you have to fork over the $$$$ or you are toast.

I would never ever consider buying software that stopped working after a license expires. It's just too risky - it means if you want to a product to be maintainable, you're tied to paying fees forever that could rise unpredictably, or the company could disappear or stop supporting it, leaving you totally screwed.
 
The one good thing Protel/Altium did when they bought Accel/PCAD was remove the dongle. I know I can carry on using PCAD for ever, the only risk being MS breaking something in a new version of Windows, but that hasn't happened yet.

[Rufus]

Might work factory wise, but software wise - from experience - it always goes tango uniform.

IMHO, AD is now dead because it is written in Delphi (!!), you cant find good Delphi guys without ages of HR

Gees not like Pascal is rocket science, if a programmer can't become productive with Delphi in a couple of weeks they are not worth having anyway.

[Rufus]

The big BUT of course is it effectively locks you into the Altium system, once that 12 month license runs out, you have to fork over the $$$$ or you are toast.

I would never ever consider buying software that stopped working after a license expires. It's just too risky

At about £260 for a nanoboard you are being given a year of software not buying it.

It is for evaluation, if you use it seriously you need to buy the full package which is perpetual excepting some Altium live features and the vault storage drivel.

[Psi]

Might work factory wise, but software wise - from experience - it always goes tango uniform.

IMHO, AD is now dead because it is written in Delphi (!!), you cant find good Delphi guys without ages of HR

Gees not like Pascal is rocket science, if a programmer can't become productive with Delphi in a couple of weeks they are not worth having anyway.

Yeah, switching to Delphi is real easy. Talking to the windows API is the same in any language.

Delphi isn't dead, it's still going strong. Although it is more popular in some countries than others.

[Greg J]

But because its universal (delphi) you are on the mercy of its designers, etc. You pretty much isolate yourself in one world.
As a software engineer for 20 years, I have to say - delphi is shite! Simple and utter crap.
Good software engineer knows at least 2 languages, and some scripting language. But for someone with experience, learning a new language is not an issue.
One that wants to stick to one language to rule them all, and thinks that this is good - should really change business, and start perhaps sweeping streets. Not much changes in that business.

[rz2k]

Gees not like Pascal is rocket science, if a programmer can't become productive with Delphi in a couple of weeks they are not worth having anyway.

to go productive with delphi you need to know all bufgixes for bugfixes for bugfixes from all versions. it is not a just pascal.
thats why people still use delphi 2007.

and heres some examples:
FL Studio - 11 years of development by 2 guys.
Guitar Pro - made by little team, in version 6 switched to C
Total Commander - made by one guy in Delphi 4 (1998), new versions still done in that version.
Skype - Delphi is only GUI frontend, all magick is in DLLs done in C

AD10 is now one and only CAD done in Delphi.

Talking to the windows API is the same in any language.

Its not about winapi, like I already said - if you lose your original team, who made the original product concept and first versions - you will need to spent BIG times with HR to find new people, who will not only "know pascal and winapi", but understand all layers of shit in Delphi and understand what you need from them.

AD10 is not a skype or total commander, AD10 must be done in 100% tested code, because, people will go nuts, if in project, they have spent, for example 2 month, they will find some cool bug, like dead PDF exporter or DRC errors from nowhere. AD is engineering soft for $5K ($10K here) which needs to remain stable all times, in Delphi, it is a HUGE task to do.

[Greg J]

I thought Skype used QT C++ GUI (very good one by the way!) libraries.

[Bloch]

Total Commander - made by one guy in Delphi 4 (1998), new versions still done in that version.

Wrong !  http://www.ghisler.ch/board/viewtopic.php?p=207611

Yes, I've been using Delphi 2 with handmade Unicode support, but I'm currently porting to Lazarus/Free Pascal which has Unicode controls based on UTF-8. Author of Total Commander

AD10 is now one and only CAD done in Delphi.
AD is engineering soft for $5K ($10K here) which needs to remain stable all times, in Delphi, it is a HUGE task to do.

Are you saying that it is only c program there are stable ? No need to test c code ?

I am sorry but will you not explain me why Programs made by a Delphi compiler are more unstable I love to learn about that. 

[EEVblog]

I am sorry but will you not explain me why Programs made by a Delphi compiler are more unstable I love to learn about that.

You could argue that because there are orders of magnitude less people using Delphi than C, the compilers may not be as well proven.
Whether or not that's the actual case, with Altium, I don't know.

Dave.

[Bloch]

You could argue that because there are orders of magnitude less people using Delphi than C, the compilers may not be as well proven.

My argument will then be that because there are more different vendors / compiler versions / dialect (c/c++/c#) and library's. It may be as easy to find an expert in Delphi than C.

And one more thing. It is so easy to make GUI in C/c++ that some need to use example MFC / QT :-) There are some sarcasm in this line :-)

[Greg J]

Now, think how you guys feel when software engineers talk about hardware design. It is that much 'fun' for me to read all the statements coming from hardware guys about software.
Sheer entertainment

[EEVblog]

My argument will then be that because there are more different vendors / compiler versions / dialect (c/c++/c#) and library's. It may be as easy to find an expert in Delphi than C.

It's not.
No programmer hired at Altium is expected to know Delphi, in fact it was very rare to find anyone who had ever used it AFAIK.
Not that it matters. It doesn't take long for a good programmer to come up to speed on it.
getting up to speed on the actual product code, now that's a different matter entirely.

Dave.

[DrGeoff]

No programmer hired at Altium is expected to know Delphi, in fact it was very rare to find anyone who had ever used it AFAIK.

The only other place I know that used Delphi was AWA Wagering.
Maybe Altium picked up some of them when it was closed down or sold off.