You can copy the *.net file while LTC is open if you want to keep it. I use a bat file in Sendto to clean up other LTC files in subdirectories....
Anyhow, I came here to say that when I overdrove Mark's model, it had terrible rail sticking at 1KHz, even. So, I wanted to see if Teemuk/Ian's model did the same. Well no, but it does have a terrible phase reversal problem on the negative clip. The problem comes from the VAS Darlington dumping current into the VAS degen resistor. Of course, this is in the LM3886 chip, so has anyone with a real chip seen this problem? I would be surprised if such a serious problem wasn't fixed, but for an accurate simulation, we would like to know just how. A current limit on the VAS turns out to be quite picky but a series resistor on the first collector is more forgiving, but it may affect the phase margin. This image shows a possible fix (Rfix) to the model, but who knows what is on the chip?
Anyhow, I came here to say that when I overdrove Mark's model, it had terrible rail sticking at 1KHz, even. So, I wanted to see if Teemuk/Ian's model did the same. Well no, but it does have a terrible phase reversal problem on the negative clip. The problem comes from the VAS Darlington dumping current into the VAS degen resistor. Of course, this is in the LM3886 chip, so has anyone with a real chip seen this problem? I would be surprised if such a serious problem wasn't fixed, but for an accurate simulation, we would like to know just how. A current limit on the VAS turns out to be quite picky but a series resistor on the first collector is more forgiving, but it may affect the phase margin. This image shows a possible fix (Rfix) to the model, but who knows what is on the chip?
Attachments
Last edited:
It's a shame that TI hasn't corrected these flaws in the 10 years since they were discovered. It shouldn't be up to users to figure out how to fix their models. 🙁
That's really not the point, but I see where you're coming from. My customer service mindset is just different, I guess.
Perhaps the only people that badly need a detailed model are those making cheap clones of the chip. Maybe TI has an accurate model, but it is only available with a Non-Disclosure Agreement. Those who are buying chips from TI can treat it as a black box, more or less complete solution. I know that data sheets often deliberately contain errors in the "equivalent schematic", usually fairly obvious to those who understand the chip. But that circuit often tells you how to bias the chip and use it in unusual ways, and a warning about the chip limitations.
The company that I used to work for resorted to using a Chinese vendor to build products that used chips only the company could make, in order to maintain control of the product. But using a Chinese manufacturer was the only way to compete in the low-price market.
The company that I used to work for resorted to using a Chinese vendor to build products that used chips only the company could make, in order to maintain control of the product. But using a Chinese manufacturer was the only way to compete in the low-price market.
Perhaps the only people who wheedle and beg and thunderously DEMAND another simulation model, are low volume hobbyists whose cumulative purchasing volume is negligible. Perhaps the high volume customers who buy 1 x 10^7 parts per year, posses the manpower and the dedicated funds+equipment to build and measure prototype boards, then make informed judgments about what modifications to make, if any. With or without simulations.
Hi Mooly,
Thanks again for taking your time to look into this,
I believe everything is as it should, .net file has the right directory and is marked as "read only", yet I'm still getting this
Error on line 70 : e:u1:* u1:copyright u1:© u1:2000 u1:linear technology corporation. all rights reserved.
Unknown parameter "technology"
Error on line 10 : v3 0 n004 v
Unknown parameter "v"
Fatal Error: u1:e*: Missing gain value
The annoying thing is, I tried it on my MacBook and it worked (sort of - it wasn't giving me errors, but it wasn't amplifying properly in the same circuit configuration).
I'm going to re-install LTspice and try again....
Attachments
I really don't know I'm afraid. I can only relate to Windows, I've never used a Mac. I just don't know 🙁
This doesn't have much to do with customer service mindset but just how running a business works.
Spending more resources (read:money) into things that won't give you any or barely any profit is just not sustainable.
Obviously there is a balance to be made there.
When it comes to TI the service they provide has been going down rapidly the last 5 years.
I personally even think we can be thankful we have some basic models at all to begin with thanks to some enthusiasts at TI.
If you look at where the bulk of their products end up. It wouldn't surprise me that the amount of their customers who even remotely care (or even know) about SPICE modeling is less than 10%
Fixing a die may mean millions of dollars. You had better sell tends of millions of the part.
No help on my over optimistic THD measures or if I am interpreting my phase margin correctly?
I don't follow what you did to get there? Attached is a zip file with working Mooly files, including a copy of the net file. Do not move any files out of the "mooly" folder.
However, the copyright line is a comment that starts with an asterisk, and LTC should ignore it, not attempt to decode it. I think the asterisk has to be the first character in the line but you did not share the file so I dunno if that rule has been abused.
Also, this reminds me that different systems handle end-of-line differently. The asterisk may not be recognized as a comment because it follows an end-of-line character that is not correct for the problem system. You might try editing the file in Notepad++ or similar text editor that supports control characters.
@Kannan.cnkr this thread is purely about the LTspice simulator. As your issue relates to a real amplifier I'll move your post to 'Chip Amp' fourm 🙂
Best to use a TIAN probe in the feedback path, there is a Ltspice example circuit that you can use, the file is named LoopGain2.asc and you can find this in (Microsoft Windows) - C:\Users\username\Documents\LTspiceXVII\examples\Educational
Last edited:
Hi - I am having trouble finding any way of increasing the frequency axis beyond 100 Khz - is this just not possible?
