lm3886 servo help

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all,

I want to draw a pcb of the 3886 using a dc servo, but I have a few questions before pulling the trigger.

I´ll be using the back plane as a power ground for power caps and zobel return, while the front will be the plane for signal ground. There will be pads on each plane so I can put a jumper or resistor in case I get hum or noise.

My question is if the ground at the DC servo net is considered as power ground, or being low current, could be put at the signal ground. R8 is signal ground too?

I plan to use a TL071 as I have some spares, but have read that the LF411 works better. Can I use both of them without changing components values?

How can I calculate values for R4, R5 and R6? I tried to find some info for that but had no luck on it. I´ll be glad if you could send me a link or explain me how to choose the right values for those.

I include the schematic. Saw this one on other thread, but added a zobel and thiele networks and a high cutoff filter at the input.

Please, if you find any error in that schematic let me know.

Thanks a lot for your help,

lm3886 servo schematic 3.jpg
 
Thanks Andrew,

according to what Tom Christiansen wrote:

"There is much confusion about the naming conventions of these output filters. Technically, the RC series network is the Zobel network and LR parallel combination a Thiele network. Occasionally, the Zobel network is referred to as a Boucherot cell as well. In the schematic below, it would have been more consistent with the naming convention to use Lt and Rt rather than Lz and Rz2. As previously stated, there’s much confusion… I will do my best to limit the confusion from here on."

I am not saying you are wrong, I only need to know if that schematic is right and I can start working on it.

I´ll be glad if someone could send me the link to the thread regarding dc servos or explain a bit about how to calculate resitor values.

Thanks!
 
Member
Joined 2009
Paid Member
Andrew, I'm perfectly willing to listen as long as you back your claims up with evidence rather than just passing them off as gospel. Just because you say something, doesn't make it right or valid.

You can have a look at Douglas Self, Bob Cordell, and Randy Slone's works and you'll find there's some debate about what the two reactive networks are called. I chose to put a stake in the ground and refer to the L||R as the Thiele Network and the R+C as the Zobel Network. Some refer to all four components as a Boucherot Cell. Others say the R+C is the Boucherot Cell. Yet others, call the R+C a snubber – and they'd be correct in their application of that term as well.

I am certainly willing to take another look at the naming conventions and the wording of my website (the Stability page needs an update anyway), but it's not a high priority at this point. Unless you provide compelling evidence for me to change my naming conventions, I'm going to leave it as-is.
The other thing to consider is what to do if the naming conventions are indeed murky and the topic of contentious debate. What should we do then? Does your word become law or can we agree to have some level of confusion and as long as we define our terms before using them, we can both coexist?

I'm perfectly willing to work with you on this. The ball's in your court.

Tom
 
Last edited:
this may be where it is shown:
Thiele, A. Neville (1975). "Load Stabilising Networks for Audio Amplifiers," Proceedings of the Institute of Radio and Electronics Engineers, 36(9), pp. 297–300. Reprinted Journal of the Audio Engineering Society, 1976, 24(1), pp. 20–23.
I am not a Member so I can't look up the paper.

E.Cherry did a paper discussing the Thiele Stabilising Network in one of the Tech magazines and did a follow up a couple of years later.
I'll try to find the link. It's in a few posts I have made years ago.
There was a Forum download of an extract from the E.Cherry paper, but it would be more difficult to find.
I'll see what I can find.

Electronics World: Jan 1995 and July 1997
Here's the excel, see sheet Zobel
It was written in a very old MS version and does not always load up correctly, but usually legible.
 

Attachments

  • electronicsrev1.xls.zip
    371 KB · Views: 42
Last edited:
this may be where it is shown:
Thiele, A. Neville (1975). "Load Stabilising Networks for Audio Amplifiers," Proceedings of the Institute of Radio and Electronics Engineers, 36(9), pp. 297–300. Reprinted Journal of the Audio Engineering Society, 1976, 24(1), pp. 20–23.
I am not a Member so I can't look up the paper.

Cited by four people, says Google Scholar. That's not exactly a sign of authoritative work. Just saying. In all fairness, it could be the reprint of the paper that's cited four times.

I'm contemplating becoming an AES member. If I do so, I'll take a look at the paper. Sadly, I don't have access to AES through the university.

E.Cherry did a paper discussing the Thiele Stabilising Network in one of the Tech magazines and did a follow up a couple of years later.
I'll try to find the link. It's in a few posts I have made years ago.

I'd appreciate it.

Rather than continuing the OT discussion here, would you just toss me a PM once you dig that up?

Tom
 

Attachments

  • Screen Shot 2017-11-19 at 3.48.14 PM.png
    Screen Shot 2017-11-19 at 3.48.14 PM.png
    41.5 KB · Views: 307
I´m sorry, but I think that thread is deviating...
I didn´t want to start a debate about how those networks are called.

I agree. Sorry about that. When I'm being thrown under the bus in public, I do feel a certain need to stand up for myself.

I need to know if the schematic I attached contains any errors or if I can start working on it.

At first glance, it looks pretty good. Note that C2 will cause overshoot on the step response. That can be fixed by adding an R+C across R3. This is shown in the LM3886 data sheet (Test Circuit #2 on page 5). I'd go with R4 = 10*R3.
You need supply connections on U3, U4. D1 is not needed if you size the mute resistor (R7) for 0.5-1.0 mA at the nominal supply voltage.

For information on how to design DC servos, have a look at this post: Is this a good LM3886 Kit? (Post #98)

Tom
 
Last edited:
Asuslover, thanks a lot for the link.

Very interesting design there. You tried it?

Tom, thanks so much for taking a look.

I thought I could use a low value cap between pins 9 and 10 without needing a RC in the feedback loop.

what about removing C2?

Yeah, I forgot the supply cons on u3, and u4. They will be connected to the power supply (+24v, -24v ,unregulated)

Regarding the zener diode... deleting it and going larger value on R7 will have the same effect? I want a "soft-on" start without having to go with regulators.

You say R4 should be 10*R3, and... R5=100*R3? right?

Thanks too for the link to your post. Very useful info there.

I think I´ll have to order some 411s to try them against the tl071.

Regards,

David
 
I thought I could use a low value cap between pins 9 and 10 without needing a RC in the feedback loop.

what about removing C2?

The cap between the inputs of the LM3886 (C2 in your schematic) prevents quasi-oscillations as the output swings towards the negative rail. What happens is that the VAS transistor in the LM3886 enters saturation and its gain drops like a rock. This causes instability and the output will buzz as it reaches the negative supply voltage. The result is slightly lower than specified output power (and spurious oscillations). To address this, C2 is added. That messes up the transient response, so Rf2 and Cf (see Test Circuit #2 in the data sheet) are added.

Regarding the zener diode... deleting it and going larger value on R7 will have the same effect? I want a "soft-on" start without having to go with regulators.

Calculating R7 so you have at least 0.5 mA at the absolute lowest value of supply voltage (low mains, bottom of the ripple valleys) will guarantee that the chip works well and starts up without any audible pops.
If you want round numbers, aim for 0.5 mA at 2/3 of the nominal supply voltage. Note that the MUTE pin sits at -3 V when the chip is un-muted, so you need to factor this in when you calculate R7. Also, round down to the nearest standard value to ensure that enough current flows.

You say R4 should be 10*R3, and... R5=100*R3? right?

R5 is determined by the cutoff frequency of the DC servo. It has no relation to R3. For the tradeoffs regarding the cutoff frequency, see the DC servo thread I linked to a few posts back.


I do. I'm also aiming for microvolt offsets. The LF411 is getting expensive. You can find some of the TI/Burr-Brown opamps (OPA...) that provide much better performance at the same price point.

If you haven't already, I suggest dropping the circuit into a spice simulator, such as TINA-TI (free download from TI). Simulate the DC performance with the various opamps used in the DC servo. That'll provide more insight faster than any lab experiment - at least if you're reasonably comfortable with spice simulation software.

Tom
 
Hi again!

Ordered some boards according to the schematic and your advices. Got it running and even sounds great,but I´m afraid the servo is not working.

I measure from ground to lf411 pin 7 and get +15V, great!! then I measure ground to pin 4 and get -30V!!! Can´t find what is wrong.

At first I thought those -30volts came directly from PS but the power supply pumps 34,5volts so I´m a bit lost.

I get the same reading for both boards so I think its not a faulty 79L15...

I draw the servo supply like at AN-1192 pdf, figure 16.

Whats the procedure to find out if the servo is working or not?

Need your help guys, I can´t find out what I did wrong.

Thanks again!
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.