Re: Re: TMC
Hi Edmond,
I think this is a good explanation. I think it is also fair to say that one can inadvertantly get wrong values in applying either TPC or TMC, and consequently get disappointing results.
Happy Holidays,
Bob
Edmond Stuart said:
Hi Edmond,
I think this is a good explanation. I think it is also fair to say that one can inadvertantly get wrong values in applying either TPC or TMC, and consequently get disappointing results.
Happy Holidays,
Bob
disappointing results
Hi Bob,
Sure, I know of at least one recent example of such erroneous measurements. 😀
Happy Holidays too!
Edmond.
Hi Bob,
Sure, I know of at least one recent example of such erroneous measurements. 😀
Happy Holidays too!
Edmond.
Bob Cordell said:
Hi Bob.
I tried that too, but the extra C at the VAS collector had a detrimental effect on the slew rate.
Cheers,
Glen
G.Kleinschmidt said:
Hi Glen,
Yes, you are right. In the experiment that I did, the slew rate was reduced to a little less than 100 V/us. I didn't work hard at optimizing it, but the point is well-taken. The original circuit, employing the series R-C across the LTP collectors, slews much more effortlessly. BTW, my VAS standing current was 10 mA.
As I mentioned before, I don't like sticking, even if it is fairly brief and benign (e.g., not the result of saturation). However there are sometimes tradeoffs. Although how an amplifier behaves in clipping I think is important, there are much worse misbehaviors in clipping in the real world that can occur than 1 us of recovery time. Slamming against dirty rails is a good example.
In fairness to the no-NFB advocates, clipping behavior is one of the things that the application of negative feedback can make worse if one is not careful. Of course, I have also seen no-NFB designs that clip horribly.
Ideally, amplifiers should never clip, but when they do, they should do so as cleanly as possible, hopefully not too much at the expense of normal performance. I suspect that power amplifiers clip more often than some believe, and have seen it with proper instrumentation on well recorded music with good dynamic range. In this regard, I like the idea of soft-clip circuits that dynamically track the available dynamic headroom, like the Klever Klipper. These circuits cause the clipping to occur OUTSIDE the NFB loop, preventing the amplifier-proper from ever clipping. Interestingly, these soft-clip approaches cause increased amounts of relatively benign odd-order distortion at a power point well below clipping, so in fact their objective distortion numbers would be degraded (kind of like a tube amp).
One could, of course, also choose to apply very abrupt clipping outside the NFB loop as well.
I have a few other ideas for reducing that small amount of sticking without compromising slew rate performance as much, and I'll let you know if they pan out.
Cheers,
Bob
Bob Cordell said:
Thanks Bob. I've finalised my design now and I'm going ahead with it but I look forward to your results.
WRT to clipping/sticking and GNFB Vs non-GNFB, this is probably one area where traditional miller compensation around the VAS scores a point in its favour as there are no sticking issues accociated with over-charging of the compensation capacitor.
WRT to clipping recovery, stability and slewrate, I'm finding it very hard to match the all-over performance of the Cdom-compensated front end I made for one of my bigger amps:
http://www.diyaudio.com/forums/showthread.php?postid=1591554#post1591554
😀 😀 . However, a degree of extra complexity was required here.
Cheers,
Glen
Hey Glen,
That other design of yours, where you tripled up on the VAS transistors - were these TO-126 devices? Or did you actually triple up on TO-220's?
I take it you are going for best transient response here, and not the (targeted, simulated) sub-ppm thing?
Edit: BTW, in your sim in the referenced post, where you had two copies of the amp, one with and one without capacitive loads, you can simplify it. You can specify the load on the VAS as {CL}, then use:
.STEP param CL list 0.01p 1n
Then there is only one instance of the circuit needed. Although I may have misinterpreted what you were doing. My eyes aren't what they used to be 🙂.
That other design of yours, where you tripled up on the VAS transistors - were these TO-126 devices? Or did you actually triple up on TO-220's?
I take it you are going for best transient response here, and not the (targeted, simulated) sub-ppm thing?
Edit: BTW, in your sim in the referenced post, where you had two copies of the amp, one with and one without capacitive loads, you can simplify it. You can specify the load on the VAS as {CL}, then use:
.STEP param CL list 0.01p 1n
Then there is only one instance of the circuit needed. Although I may have misinterpreted what you were doing. My eyes aren't what they used to be 🙂.
andy_c said:
Hi Andy.
Thanks for the LTspice tip. That will make things a little easier in the future 🙂
The VAS transistors are TO-126 (those KSC/KSA ones - recently bought 250pc of each from Mouser for peanuts).
That front end (amongst ten dozen other things I'm currently loading the final interation of the PCB's) still has very low distortion (not dominant over the output stage). I've down graded the power rating though; now I'm only using 16 pairs of output deviced with another 16 pairs in the series pass power supply. Bias current is 5A and max output current is limited to 80A.
Distortion of this output stage into anything but a very low impedance load is very low too.
Currently waiting on delivery of the transformers and filter capacitors.
Cheers,
Glen
andy_c said:
Are they? LOL!.
Don't try to buy any obsolete U404 dual JFET from Farnell either 😀
Cheers,
Glen
IP ban
Perhaps one may wonder why I wrote this: 'does work ...', implying that the mirror site (http://www.synaesthesia.ca) of the PGP amp is down.
At first, I thought it was only temporarily closed due to maintenance. However, that wasn't the case. It turned out that my IP address has been blocked! What a childish action. Even a 'pitchfork wielding villager' doesn't do such thing. How deep can you fall, syn08?!
Edmond Stuart said:
Perhaps one may wonder why I wrote this: 'does work ...', implying that the mirror site (http://www.synaesthesia.ca) of the PGP amp is down.
At first, I thought it was only temporarily closed due to maintenance. However, that wasn't the case. It turned out that my IP address has been blocked! What a childish action. Even a 'pitchfork wielding villager' doesn't do such thing. How deep can you fall, syn08?!
dimitri said:
That's funny 😀
I took a look at Edmond's complain - I would suggest checking with the ISP. I had the same problem with a few other visitors, described here http://www.diyaudio.com/forums/showthread.php?postid=1703864#post1703864 I also recall a similar situation a year+ ago when the email communication from/to tiscali.nl was broken. It was a DNS issue when tiscali.nl changed ownership.
Certainly, I have no technical means to filter the IPs allowed to visit my site.
syn08 said:
You did what?
You promised me to put me on your ignore list. 😀That's also the reason why I thought you had blocked my IP address as well. That looks plausible because on other PCs or via a proxy, I could reach your website without any problem.
BTW1, I did a 'flushdns' without success. My next step will be checking whether other people, also connected to tiscali/telfort, have the same issues.
BTW2, You can't upload to the tiscali-PGP site any longer, because I have changed the password. I had to do that, as it uses the same PW for my e-mail as well. As you will certainly understand, I don't like the idea that you or anybody else has access (in principle) to my email account.
IP ban???
A friend of mine, having the same provider and living in the same city, experienced no problems. How's that possible?
Edmond Stuart said:
A friend of mine, having the same provider and living in the same city, experienced no problems. How's that possible?
Edmond,
1. "ipconfig /release" then "ipconfig /renew" and see if it's working
2. Check the "hosts" file (no extension) in C:\<windows dir>\system32\drivers\etc for any stray entries. It should have only an
127.0.0.1 localhost
entry.
3. "tracert www.synaesthesia.ca" See if the tracert gets stuck somewhere (routing issue) or it rejects right away with "destination host unknown" (DNS issue).
4. "ipconfig /all" and compare the settings with your friend's (in particular the DNS settings).
1. "ipconfig /release" then "ipconfig /renew" and see if it's working
2. Check the "hosts" file (no extension) in C:\<windows dir>\system32\drivers\etc for any stray entries. It should have only an
127.0.0.1 localhost
entry.
3. "tracert www.synaesthesia.ca" See if the tracert gets stuck somewhere (routing issue) or it rejects right away with "destination host unknown" (DNS issue).
4. "ipconfig /all" and compare the settings with your friend's (in particular the DNS settings).
syn08 said:
Done.
Done, is OK.
If routed via: 'Netherland, Amsterdam, Support Net'
I got 'Not Found'. Also when using your IP address (66.6.63.32)
But if routed via 'Greece, Athens, NTUA - National Technical University of Athens' (for example) I got no error msg.
So, ''Netherland, Amsterdam, Support Net' has issues. But why?
Hmm... that's not so easy, as she hasn't the faintest idea how to do that, let alone to ask her to make screen-shots and send them to me. Nevertheless, I think they are exactly the same, because I have installed the S/W etc. on her PC. Moreover, the settings of my two PCs are pretty standard and I have recently reinstalled Windows.
Before and after these re-installations I had issues with your website, so I think that 'Netherland, Amsterdam, Support Net' is the real culprit. It's not a DNS error, as 66.6.63.32 doesn't work either.
Any idea how to proceed?