Thanks! The step response and square wave response look very clean.
I had some initial trouble with a vicious spike on the leading edges from the second square wave onwards for higher signal levels, essentially going to full rail voltage and then coming down from there in a mess. A non-oscillating mess, but still a mess. At the same time, a single step response just showed a small overshoot. I eventually realized that the leading edge of my square waves were way too steep for the circuit, hence the reduced input filter frequency. It does not eliminate the problem, but prevents it from being triggered.
I get phase margins around 60-65 degrees both for the inner loop alone and the outer loop alone using the cut-the-loop-and-duplicate-the-circuit technique, but I am not sure what the graphs actually mean when I cut both nested loops where they coincide and measure the loop gain there. I get 53 degrees phase margin at the zero loop gain point, but essentially zero apparent phase margin through the entire audio band before it pulls up to 50 degrees or so around 1 MHz. It seems to have around 72 dB combined NFB at 20 kHz with 20 dB overall gain, some of that gain in the differential input buffer.
My guess is that a nested loop is a very different animal from the ordinary second order loop. There does not seem to be any ill effect from the apparent zero phase margin in the audio band. Overshoot and slight oscillation on square waves does not start before the phase margin at zero loop gain approaches zero.
If I actually build this thing, I would certainly approach it with some care. Even if it is remarkably docile in simulation, I get the sense that it could easily blow up in some unexpected way. The potential for spiking to full rail voltage is certainly lurking in there. For now, it is an interesting exercise, but I'll start a separate thread if I take it further than that.
I had some initial trouble with a vicious spike on the leading edges from the second square wave onwards for higher signal levels, essentially going to full rail voltage and then coming down from there in a mess. A non-oscillating mess, but still a mess. At the same time, a single step response just showed a small overshoot. I eventually realized that the leading edge of my square waves were way too steep for the circuit, hence the reduced input filter frequency. It does not eliminate the problem, but prevents it from being triggered.
I get phase margins around 60-65 degrees both for the inner loop alone and the outer loop alone using the cut-the-loop-and-duplicate-the-circuit technique, but I am not sure what the graphs actually mean when I cut both nested loops where they coincide and measure the loop gain there. I get 53 degrees phase margin at the zero loop gain point, but essentially zero apparent phase margin through the entire audio band before it pulls up to 50 degrees or so around 1 MHz. It seems to have around 72 dB combined NFB at 20 kHz with 20 dB overall gain, some of that gain in the differential input buffer.
My guess is that a nested loop is a very different animal from the ordinary second order loop. There does not seem to be any ill effect from the apparent zero phase margin in the audio band. Overshoot and slight oscillation on square waves does not start before the phase margin at zero loop gain approaches zero.
If I actually build this thing, I would certainly approach it with some care. Even if it is remarkably docile in simulation, I get the sense that it could easily blow up in some unexpected way. The potential for spiking to full rail voltage is certainly lurking in there. For now, it is an interesting exercise, but I'll start a separate thread if I take it further than that.
Last edited:
I tried noise gain compensation on the LM3886. That was a dead end. It works great for the phase margin, but it results in a little overshoot. No big deal you think, until you crank up the signal level and that overshoot turns into a step. When you build the circuit, you'll find that at that signal level, the circuit will oscillate. It's one of those operating corners that isn't covered by the LM3886 model.
Tom
Tom
As some of you may have noticed, I've started a group buy for the absolute last board run on the Parallel-86. If you are interested, please pre-order your boards here: Parallel-86 (Last Chance to Buy)
I expect to reach the minimum of ten boards pre-ordered by next weekend (July 23/24) so I can order boards on the 25th. This will be the absolutely final run.
RS and Newark/Farnell sit on the world-wide distributor stock and it's all in the UK. There does seem to be quite a few people who have a couple of LM4780s rattling around in desk drawers, though. Please ensure that you have a source of the LM4780 before you order.
Tom
I expect to reach the minimum of ten boards pre-ordered by next weekend (July 23/24) so I can order boards on the 25th. This will be the absolutely final run.
RS and Newark/Farnell sit on the world-wide distributor stock and it's all in the UK. There does seem to be quite a few people who have a couple of LM4780s rattling around in desk drawers, though. Please ensure that you have a source of the LM4780 before you order.
Tom
I've committed to Fremen that if he sends me boards, I'll have the boards (or at least a board) assembled by the end of summer (so early September). He requested that I followed the build instructions. I will do that. It is a completely reasonable request. No worries there.
I'm curious as well. Ahhh..... Data. Useful for making design decisions.
Cheers.
Tom
Tom,
Any progress on this (the above quote is from March 2016)? Did Fremen send you his boards? I for one am very interested in the measurements
Be well and have a profitable summer
Anand.
Last edited:
Any progress on this (the above quote is from March 2016)? Did Fremen send you his boards? I for one am very interested in the measurements
I have the MyRef boards. Thanks Dario. They make fantastic coasters on my desk right now...
(Sorry, Dario).The boards are actually pretty nice. 2.0 mm (?!) thick fibreglass. ENIG finish. Tweakers will be delighted to find both holes for leaded components and pads for SMD components. Unfortunately this also means that the board is optimized for neither. The boards are not bad. There's some room for improvement and I know Dario is working to continue to improve the layout.
I have Rev. 1.6 boards. The latest rev is now 1.7. If I was in Dario's shoes, I'd set up a prototyping board fab in my garage. It's so much easier to tweak the board when you can see the results in a couple of hours rather than waiting a couple of weeks for your boards to arrive in the mail. It took me a few (two I think) garage fab iterations of the layout to arrive at the Modulus-86 Rev. 1.0 and it spun off the LM3886 P2P vs PCB thread in the process. Then again, I'm more of a do-it-once-and-be-done kinda guy than a tweaker. Each to his own.
I should touch base with Dario to see if measurements are still relevant. I'd hate to end up in a situation where I spend valuable time and money characterizing the circuit only to have the results invalidated by "but, but... you didn't measure the latest rev". If he says it's worth it, I'll proceed. It'll have to wait a bit. Right now I need to push on activities that will make me money rather than charity activities. That's the unfortunate reality for those of us who are self-employed.
Be well and have a profitable summer
That's the plan. I hope to spend my summer getting set up well for the fall so I can focus on my studies then.
I would love to have a few bits for show & tell at the Head-Fi meet-up in Calgary this September and have two other major projects as well.
Tom
Last edited:
You're welcome.
They make fantastic coasters on my desk right now...
They should have a better use...
If you have any suggestion I will be happy to look at and appreciate it.
Revisions make small changes to fix small glitches, revision 1.7 is likely to be stable for a long time, like it was 1.02.
And BTW I don't have a garage... I live in small studio apartment, so no room for it
You will not hear any "but, but... you didn't measure the latest rev" from me
I already know that when the complete amp will be measured it will fare well but a bit less than JosephK measurements since he did it with the input cap replaced by a wire and bypassing the DC protection circutry.
Both can introduce some noise and so increase THD+N.
What counts is how good it sounds and the much lower distortion figures tell me I've did a good job optimizing the layout.
There's room for improvement, obviously.
1.7 version simply adjust some markings, move C9, ditches C21 and can host a bit bigger input cap.
I don't expect it to measure or sound different than 1.6.
I'm not in a hurry but don't make that bit a lot...
I wholeheartedly agree.
And BTW I don't have a garage... I live in small studio apartment, so no room for it
When I lived in a studio apartment, I had the prototype fab set up on my kitchen counter and performed all the work there ... including applying photo resist, exposure, development (in lye (NaOH)), and etch in ferric chloride. I somehow managed to not make a mess and never get ferric chloride on the carpet. For drilling the holes, I had access to an actual PCB drilling station in college. Before then I used my dad's drill stand and power drill. Now I use a drill press ... in the garage.
On the other hand, there are prototyping PCB manufacturers out there who are incredibly inexpensive so it almost doesn't make sense to run a garage fab anymore.
Ah. Good to know.
I'm not in a hurry but don't make that bit a lot...
I figure early October. Hopefully sooner.
Tom
Well, I ordered the parts. So I now have a bag of parts and boards. So progress has been made. I do have to focus on paying projects before charity work. Without money I don't get to eat...
Sorry. Reality bites. I will get the circuit measured one of these days. I need three of my own projects out of the way first, however.
Tom
Sorry. Reality bites. I will get the circuit measured one of these days. I need three of my own projects out of the way first, however.
Tom
I asked in post 122 about Thiele/Zobel for a Sympatico (Bridged 3886)
As I said,
I don't have easy access to the ground, and so am wondering whether it would be ok to do a Zobel just from + to -, and any suggested values would be helpful.
Thanks!
As I said,
I don't have easy access to the ground, and so am wondering whether it would be ok to do a Zobel just from + to -, and any suggested values would be helpful.
Thanks!
I realized PA04 non-inverting offset control TL061 ... it works and is stable; hell the selection of 3886 amounted to offset.
http://www.diyaudio.com/forums/chip-amps/79303-chip-amp-photo-gallery-304.html
post#3040
http://www.diyaudio.com/forums/chip-amps/79303-chip-amp-photo-gallery-304.html
post#3040
I haven't had a chance to solder it up yet. I have the parts and the PCBs. I'm sorry it's dragging on. I should build that amp so I can rid myself of the guilt I feel every time I walk past the box of parts.
Tom
I'm curious too to see these measurements, and of the Sympatico as well, but he doesn't have the boards (nor the time I guess).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.